1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ********************************************************************** |
5 | * Copyright (C) 2002-2016, International Business Machines |
6 | * Corporation and others. All Rights Reserved. |
7 | ********************************************************************** |
8 | * file name: ucnv_u7.c |
9 | * encoding: UTF-8 |
10 | * tab size: 8 (not used) |
11 | * indentation:4 |
12 | * |
13 | * created on: 2002jul01 |
14 | * created by: Markus W. Scherer |
15 | * |
16 | * UTF-7 converter implementation. Used to be in ucnv_utf.c. |
17 | */ |
18 | |
19 | #include "unicode/utypes.h" |
20 | |
21 | #if !UCONFIG_NO_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION |
22 | |
23 | #include "cmemory.h" |
24 | #include "unicode/ucnv.h" |
25 | #include "ucnv_bld.h" |
26 | #include "ucnv_cnv.h" |
27 | #include "uassert.h" |
28 | |
29 | /* UTF-7 -------------------------------------------------------------------- */ |
30 | |
31 | /* |
32 | * UTF-7 is a stateful encoding of Unicode. |
33 | * It is defined in RFC 2152. (http://www.ietf.org/rfc/rfc2152.txt) |
34 | * It was intended for use in Internet email systems, using in its bytewise |
35 | * encoding only a subset of 7-bit US-ASCII. |
36 | * UTF-7 is deprecated in favor of UTF-8/16/32 and SCSU, but still |
37 | * occasionally used. |
38 | * |
39 | * For converting Unicode to UTF-7, the RFC allows to encode some US-ASCII |
40 | * characters directly or in base64. Especially, the characters in set O |
41 | * as defined in the RFC (see below) may be encoded directly but are not |
42 | * allowed in, e.g., email headers. |
43 | * By default, the ICU UTF-7 converter encodes set O directly. |
44 | * By choosing the option "version=1", set O will be escaped instead. |
45 | * For example: |
46 | * utf7Converter=ucnv_open("UTF-7,version=1"); |
47 | * |
48 | * For details about email headers see RFC 2047. |
49 | */ |
50 | |
51 | /* |
52 | * Tests for US-ASCII characters belonging to character classes |
53 | * defined in UTF-7. |
54 | * |
55 | * Set D (directly encoded characters) consists of the following |
56 | * characters: the upper and lower case letters A through Z |
57 | * and a through z, the 10 digits 0-9, and the following nine special |
58 | * characters (note that "+" and "=" are omitted): |
59 | * '(),-./:? |
60 | * |
61 | * Set O (optional direct characters) consists of the following |
62 | * characters (note that "\" and "~" are omitted): |
63 | * !"#$%&*;<=>@[]^_`{|} |
64 | * |
65 | * According to the rules in RFC 2152, the byte values for the following |
66 | * US-ASCII characters are not used in UTF-7 and are therefore illegal: |
67 | * - all C0 control codes except for CR LF TAB |
68 | * - BACKSLASH |
69 | * - TILDE |
70 | * - DEL |
71 | * - all codes beyond US-ASCII, i.e. all >127 |
72 | */ |
73 | #define inSetD(c) \ |
74 | ((uint8_t)((c)-97)<26 || (uint8_t)((c)-65)<26 || /* letters */ \ |
75 | (uint8_t)((c)-48)<10 || /* digits */ \ |
76 | (uint8_t)((c)-39)<3 || /* '() */ \ |
77 | (uint8_t)((c)-44)<4 || /* ,-./ */ \ |
78 | (c)==58 || (c)==63 /* :? */ \ |
79 | ) |
80 | |
81 | #define inSetO(c) \ |
82 | ((uint8_t)((c)-33)<6 || /* !"#$%& */ \ |
83 | (uint8_t)((c)-59)<4 || /* ;<=> */ \ |
84 | (uint8_t)((c)-93)<4 || /* ]^_` */ \ |
85 | (uint8_t)((c)-123)<3 || /* {|} */ \ |
86 | (c)==42 || (c)==64 || (c)==91 /* *@[ */ \ |
87 | ) |
88 | |
89 | #define isCRLFTAB(c) ((c)==13 || (c)==10 || (c)==9) |
90 | #define isCRLFSPTAB(c) ((c)==32 || (c)==13 || (c)==10 || (c)==9) |
91 | |
92 | #define PLUS 43 |
93 | #define MINUS 45 |
94 | #define BACKSLASH 92 |
95 | #define TILDE 126 |
96 | |
97 | /* legal byte values: all US-ASCII graphic characters from space to before tilde, and CR LF TAB */ |
98 | #define isLegalUTF7(c) (((uint8_t)((c)-32)<94 && (c)!=BACKSLASH) || isCRLFTAB(c)) |
99 | |
100 | /* encode directly sets D and O and CR LF SP TAB */ |
101 | static const UBool encodeDirectlyMaximum[128]={ |
102 | /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */ |
103 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, |
104 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
105 | |
106 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, |
107 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
108 | |
109 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
110 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, |
111 | |
112 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
113 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0 |
114 | }; |
115 | |
116 | /* encode directly set D and CR LF SP TAB but not set O */ |
117 | static const UBool encodeDirectlyRestricted[128]={ |
118 | /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */ |
119 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, |
120 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
121 | |
122 | 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, |
123 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, |
124 | |
125 | 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
126 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, |
127 | |
128 | 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
129 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 |
130 | }; |
131 | |
132 | static const uint8_t |
133 | toBase64[64]={ |
134 | /* A-Z */ |
135 | 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, |
136 | 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, |
137 | /* a-z */ |
138 | 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, |
139 | 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, |
140 | /* 0-9 */ |
141 | 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, |
142 | /* +/ */ |
143 | 43, 47 |
144 | }; |
145 | |
146 | static const int8_t |
147 | fromBase64[128]={ |
148 | /* C0 controls, -1 for legal ones (CR LF TAB), -3 for illegal ones */ |
149 | -3, -3, -3, -3, -3, -3, -3, -3, -3, -1, -1, -3, -3, -1, -3, -3, |
150 | -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, -3, |
151 | |
152 | /* general punctuation with + and / and a special value (-2) for - */ |
153 | -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 62, -1, -2, -1, 63, |
154 | /* digits */ |
155 | 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, -1, -1, -1, -1, -1, -1, |
156 | |
157 | /* A-Z */ |
158 | -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, |
159 | 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, -1, -3, -1, -1, -1, |
160 | |
161 | /* a-z */ |
162 | -1, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, |
163 | 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, -1, -1, -1, -3, -3 |
164 | }; |
165 | |
166 | /* |
167 | * converter status values: |
168 | * |
169 | * toUnicodeStatus: |
170 | * 24 inDirectMode (boolean) |
171 | * 23..16 base64Counter (-1..7) |
172 | * 15..0 bits (up to 14 bits incoming base64) |
173 | * |
174 | * fromUnicodeStatus: |
175 | * 31..28 version (0: set O direct 1: set O escaped) |
176 | * 24 inDirectMode (boolean) |
177 | * 23..16 base64Counter (0..2) |
178 | * 7..0 bits (6 bits outgoing base64) |
179 | * |
180 | */ |
181 | |
182 | U_CDECL_BEGIN |
183 | static void U_CALLCONV |
184 | _UTF7Reset(UConverter *cnv, UConverterResetChoice choice) { |
185 | if(choice<=UCNV_RESET_TO_UNICODE) { |
186 | /* reset toUnicode */ |
187 | cnv->toUnicodeStatus=0x1000000; /* inDirectMode=true */ |
188 | cnv->toULength=0; |
189 | } |
190 | if(choice!=UCNV_RESET_TO_UNICODE) { |
191 | /* reset fromUnicode */ |
192 | cnv->fromUnicodeStatus=(cnv->fromUnicodeStatus&0xf0000000)|0x1000000; /* keep version, inDirectMode=true */ |
193 | } |
194 | } |
195 | |
196 | static void U_CALLCONV |
197 | _UTF7Open(UConverter *cnv, |
198 | UConverterLoadArgs *pArgs, |
199 | UErrorCode *pErrorCode) { |
200 | (void)pArgs; |
201 | if(UCNV_GET_VERSION(cnv)<=1) { |
202 | /* TODO(markus): Should just use cnv->options rather than copying the version number. */ |
203 | cnv->fromUnicodeStatus=UCNV_GET_VERSION(cnv)<<28; |
204 | _UTF7Reset(cnv, UCNV_RESET_BOTH); |
205 | } else { |
206 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; |
207 | } |
208 | } |
209 | |
210 | static void U_CALLCONV |
211 | _UTF7ToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, |
212 | UErrorCode *pErrorCode) { |
213 | UConverter *cnv; |
214 | const uint8_t *source, *sourceLimit; |
215 | char16_t *target; |
216 | const char16_t *targetLimit; |
217 | int32_t *offsets; |
218 | |
219 | uint8_t *bytes; |
220 | uint8_t byteIndex; |
221 | |
222 | int32_t length, targetCapacity; |
223 | |
224 | /* UTF-7 state */ |
225 | uint16_t bits; |
226 | int8_t base64Counter; |
227 | UBool inDirectMode; |
228 | |
229 | int8_t base64Value; |
230 | |
231 | int32_t sourceIndex, nextSourceIndex; |
232 | |
233 | uint8_t b; |
234 | /* set up the local pointers */ |
235 | cnv=pArgs->converter; |
236 | |
237 | source=(const uint8_t *)pArgs->source; |
238 | sourceLimit=(const uint8_t *)pArgs->sourceLimit; |
239 | target=pArgs->target; |
240 | targetLimit=pArgs->targetLimit; |
241 | offsets=pArgs->offsets; |
242 | /* get the state machine state */ |
243 | { |
244 | uint32_t status=cnv->toUnicodeStatus; |
245 | inDirectMode=(UBool)((status>>24)&1); |
246 | base64Counter=(int8_t)(status>>16); |
247 | bits=(uint16_t)status; |
248 | } |
249 | bytes=cnv->toUBytes; |
250 | byteIndex=cnv->toULength; |
251 | |
252 | /* sourceIndex=-1 if the current character began in the previous buffer */ |
253 | sourceIndex=byteIndex==0 ? 0 : -1; |
254 | nextSourceIndex=0; |
255 | |
256 | if(inDirectMode) { |
257 | directMode: |
258 | /* |
259 | * In Direct Mode, most US-ASCII characters are encoded directly, i.e., |
260 | * with their US-ASCII byte values. |
261 | * Backslash and Tilde and most control characters are not allowed in UTF-7. |
262 | * A plus sign starts Unicode (or "escape") Mode. |
263 | * |
264 | * In Direct Mode, only the sourceIndex is used. |
265 | */ |
266 | byteIndex=0; |
267 | length=(int32_t)(sourceLimit-source); |
268 | targetCapacity=(int32_t)(targetLimit-target); |
269 | if(length>targetCapacity) { |
270 | length=targetCapacity; |
271 | } |
272 | while(length>0) { |
273 | b=*source++; |
274 | if(!isLegalUTF7(b)) { |
275 | /* illegal */ |
276 | bytes[0]=b; |
277 | byteIndex=1; |
278 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
279 | break; |
280 | } else if(b!=PLUS) { |
281 | /* write directly encoded character */ |
282 | *target++=b; |
283 | if(offsets!=nullptr) { |
284 | *offsets++=sourceIndex++; |
285 | } |
286 | } else /* PLUS */ { |
287 | /* switch to Unicode mode */ |
288 | nextSourceIndex=++sourceIndex; |
289 | inDirectMode=false; |
290 | byteIndex=0; |
291 | bits=0; |
292 | base64Counter=-1; |
293 | goto unicodeMode; |
294 | } |
295 | --length; |
296 | } |
297 | if(source<sourceLimit && target>=targetLimit) { |
298 | /* target is full */ |
299 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
300 | } |
301 | } else { |
302 | unicodeMode: |
303 | /* |
304 | * In Unicode (or "escape") Mode, UTF-16BE is base64-encoded. |
305 | * The base64 sequence ends with any character that is not in the base64 alphabet. |
306 | * A terminating minus sign is consumed. |
307 | * |
308 | * In Unicode Mode, the sourceIndex has the index to the start of the current |
309 | * base64 bytes, while nextSourceIndex is precisely parallel to source, |
310 | * keeping the index to the following byte. |
311 | * Note that in 2 out of 3 cases, UChars overlap within a base64 byte. |
312 | */ |
313 | while(source<sourceLimit) { |
314 | if(target<targetLimit) { |
315 | bytes[byteIndex++]=b=*source++; |
316 | ++nextSourceIndex; |
317 | base64Value = -3; /* initialize as illegal */ |
318 | if(b>=126 || (base64Value=fromBase64[b])==-3 || base64Value==-1) { |
319 | /* either |
320 | * base64Value==-1 for any legal character except base64 and minus sign, or |
321 | * base64Value==-3 for illegal characters: |
322 | * 1. In either case, leave Unicode mode. |
323 | * 2.1. If we ended with an incomplete char16_t or none after the +, then |
324 | * generate an error for the preceding erroneous sequence and deal with |
325 | * the current (possibly illegal) character next time through. |
326 | * 2.2. Else the current char comes after a complete char16_t, which was already |
327 | * pushed to the output buf, so: |
328 | * 2.2.1. If the current char is legal, just save it for processing next time. |
329 | * It may be for example, a plus which we need to deal with in direct mode. |
330 | * 2.2.2. Else if the current char is illegal, we might as well deal with it here. |
331 | */ |
332 | inDirectMode=true; |
333 | if(base64Counter==-1) { |
334 | /* illegal: + immediately followed by something other than base64 or minus sign */ |
335 | /* include the plus sign in the reported sequence, but not the subsequent char */ |
336 | --source; |
337 | bytes[0]=PLUS; |
338 | byteIndex=1; |
339 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
340 | break; |
341 | } else if(bits!=0) { |
342 | /* bits are illegally left over, a char16_t is incomplete */ |
343 | /* don't include current char (legal or illegal) in error seq */ |
344 | --source; |
345 | --byteIndex; |
346 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
347 | break; |
348 | } else { |
349 | /* previous char16_t was complete */ |
350 | if(base64Value==-3) { |
351 | /* current character is illegal, deal with it here */ |
352 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
353 | break; |
354 | } else { |
355 | /* un-read the current character in case it is a plus sign */ |
356 | --source; |
357 | sourceIndex=nextSourceIndex-1; |
358 | goto directMode; |
359 | } |
360 | } |
361 | } else if(base64Value>=0) { |
362 | /* collect base64 bytes into UChars */ |
363 | switch(base64Counter) { |
364 | case -1: /* -1 is immediately after the + */ |
365 | case 0: |
366 | bits=base64Value; |
367 | base64Counter=1; |
368 | break; |
369 | case 1: |
370 | case 3: |
371 | case 4: |
372 | case 6: |
373 | bits=(uint16_t)((bits<<6)|base64Value); |
374 | ++base64Counter; |
375 | break; |
376 | case 2: |
377 | *target++=(char16_t)((bits<<4)|(base64Value>>2)); |
378 | if(offsets!=nullptr) { |
379 | *offsets++=sourceIndex; |
380 | sourceIndex=nextSourceIndex-1; |
381 | } |
382 | bytes[0]=b; /* keep this byte in case an error occurs */ |
383 | byteIndex=1; |
384 | bits=(uint16_t)(base64Value&3); |
385 | base64Counter=3; |
386 | break; |
387 | case 5: |
388 | *target++=(char16_t)((bits<<2)|(base64Value>>4)); |
389 | if(offsets!=nullptr) { |
390 | *offsets++=sourceIndex; |
391 | sourceIndex=nextSourceIndex-1; |
392 | } |
393 | bytes[0]=b; /* keep this byte in case an error occurs */ |
394 | byteIndex=1; |
395 | bits=(uint16_t)(base64Value&15); |
396 | base64Counter=6; |
397 | break; |
398 | case 7: |
399 | *target++=(char16_t)((bits<<6)|base64Value); |
400 | if(offsets!=nullptr) { |
401 | *offsets++=sourceIndex; |
402 | sourceIndex=nextSourceIndex; |
403 | } |
404 | byteIndex=0; |
405 | bits=0; |
406 | base64Counter=0; |
407 | break; |
408 | default: |
409 | /* will never occur */ |
410 | break; |
411 | } |
412 | } else /*base64Value==-2*/ { |
413 | /* minus sign terminates the base64 sequence */ |
414 | inDirectMode=true; |
415 | if(base64Counter==-1) { |
416 | /* +- i.e. a minus immediately following a plus */ |
417 | *target++=PLUS; |
418 | if(offsets!=nullptr) { |
419 | *offsets++=sourceIndex-1; |
420 | } |
421 | } else { |
422 | /* absorb the minus and leave the Unicode Mode */ |
423 | if(bits!=0) { |
424 | /* bits are illegally left over, a char16_t is incomplete */ |
425 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
426 | break; |
427 | } |
428 | } |
429 | sourceIndex=nextSourceIndex; |
430 | goto directMode; |
431 | } |
432 | } else { |
433 | /* target is full */ |
434 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
435 | break; |
436 | } |
437 | } |
438 | } |
439 | |
440 | if(U_SUCCESS(*pErrorCode) && pArgs->flush && source==sourceLimit && bits==0) { |
441 | /* |
442 | * if we are in Unicode mode, then the byteIndex might not be 0, |
443 | * but that is ok if bits==0 |
444 | * -> we set byteIndex=0 at the end of the stream to avoid a truncated error |
445 | * (not true for IMAP-mailbox-name where we must end in direct mode) |
446 | */ |
447 | byteIndex=0; |
448 | } |
449 | |
450 | /* set the converter state back into UConverter */ |
451 | cnv->toUnicodeStatus=((uint32_t)inDirectMode<<24)|((uint32_t)((uint8_t)base64Counter)<<16)|(uint32_t)bits; |
452 | cnv->toULength=byteIndex; |
453 | |
454 | /* write back the updated pointers */ |
455 | pArgs->source=(const char *)source; |
456 | pArgs->target=target; |
457 | pArgs->offsets=offsets; |
458 | return; |
459 | } |
460 | |
461 | static void U_CALLCONV |
462 | _UTF7FromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, |
463 | UErrorCode *pErrorCode) { |
464 | UConverter *cnv; |
465 | const char16_t *source, *sourceLimit; |
466 | uint8_t *target, *targetLimit; |
467 | int32_t *offsets; |
468 | |
469 | int32_t length, targetCapacity, sourceIndex; |
470 | char16_t c; |
471 | |
472 | /* UTF-7 state */ |
473 | const UBool *encodeDirectly; |
474 | uint8_t bits; |
475 | int8_t base64Counter; |
476 | UBool inDirectMode; |
477 | |
478 | /* set up the local pointers */ |
479 | cnv=pArgs->converter; |
480 | |
481 | /* set up the local pointers */ |
482 | source=pArgs->source; |
483 | sourceLimit=pArgs->sourceLimit; |
484 | target=(uint8_t *)pArgs->target; |
485 | targetLimit=(uint8_t *)pArgs->targetLimit; |
486 | offsets=pArgs->offsets; |
487 | |
488 | /* get the state machine state */ |
489 | { |
490 | uint32_t status=cnv->fromUnicodeStatus; |
491 | encodeDirectly= status<0x10000000 ? encodeDirectlyMaximum : encodeDirectlyRestricted; |
492 | inDirectMode=(UBool)((status>>24)&1); |
493 | base64Counter=(int8_t)(status>>16); |
494 | bits=(uint8_t)status; |
495 | U_ASSERT(bits<=UPRV_LENGTHOF(toBase64)); |
496 | } |
497 | |
498 | /* UTF-7 always encodes UTF-16 code units, therefore we need only a simple sourceIndex */ |
499 | sourceIndex=0; |
500 | |
501 | if(inDirectMode) { |
502 | directMode: |
503 | length=(int32_t)(sourceLimit-source); |
504 | targetCapacity=(int32_t)(targetLimit-target); |
505 | if(length>targetCapacity) { |
506 | length=targetCapacity; |
507 | } |
508 | while(length>0) { |
509 | c=*source++; |
510 | /* currently always encode CR LF SP TAB directly */ |
511 | if(c<=127 && encodeDirectly[c]) { |
512 | /* encode directly */ |
513 | *target++=(uint8_t)c; |
514 | if(offsets!=nullptr) { |
515 | *offsets++=sourceIndex++; |
516 | } |
517 | } else if(c==PLUS) { |
518 | /* output +- for + */ |
519 | *target++=PLUS; |
520 | if(target<targetLimit) { |
521 | *target++=MINUS; |
522 | if(offsets!=nullptr) { |
523 | *offsets++=sourceIndex; |
524 | *offsets++=sourceIndex++; |
525 | } |
526 | /* realign length and targetCapacity */ |
527 | goto directMode; |
528 | } else { |
529 | if(offsets!=nullptr) { |
530 | *offsets++=sourceIndex++; |
531 | } |
532 | cnv->charErrorBuffer[0]=MINUS; |
533 | cnv->charErrorBufferLength=1; |
534 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
535 | break; |
536 | } |
537 | } else { |
538 | /* un-read this character and switch to Unicode Mode */ |
539 | --source; |
540 | *target++=PLUS; |
541 | if(offsets!=nullptr) { |
542 | *offsets++=sourceIndex; |
543 | } |
544 | inDirectMode=false; |
545 | base64Counter=0; |
546 | goto unicodeMode; |
547 | } |
548 | --length; |
549 | } |
550 | if(source<sourceLimit && target>=targetLimit) { |
551 | /* target is full */ |
552 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
553 | } |
554 | } else { |
555 | unicodeMode: |
556 | while(source<sourceLimit) { |
557 | if(target<targetLimit) { |
558 | c=*source++; |
559 | if(c<=127 && encodeDirectly[c]) { |
560 | /* encode directly */ |
561 | inDirectMode=true; |
562 | |
563 | /* trick: back out this character to make this easier */ |
564 | --source; |
565 | |
566 | /* terminate the base64 sequence */ |
567 | if(base64Counter!=0) { |
568 | /* write remaining bits for the previous character */ |
569 | *target++=toBase64[bits]; |
570 | if(offsets!=nullptr) { |
571 | *offsets++=sourceIndex-1; |
572 | } |
573 | } |
574 | if(fromBase64[c]!=-1) { |
575 | /* need to terminate with a minus */ |
576 | if(target<targetLimit) { |
577 | *target++=MINUS; |
578 | if(offsets!=nullptr) { |
579 | *offsets++=sourceIndex-1; |
580 | } |
581 | } else { |
582 | cnv->charErrorBuffer[0]=MINUS; |
583 | cnv->charErrorBufferLength=1; |
584 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
585 | break; |
586 | } |
587 | } |
588 | goto directMode; |
589 | } else { |
590 | /* |
591 | * base64 this character: |
592 | * Output 2 or 3 base64 bytes for the remaining bits of the previous character |
593 | * and the bits of this character, each implicitly in UTF-16BE. |
594 | * |
595 | * Here, bits is an 8-bit variable because only 6 bits need to be kept from one |
596 | * character to the next. The actual 2 or 4 bits are shifted to the left edge |
597 | * of the 6-bits field 5..0 to make the termination of the base64 sequence easier. |
598 | */ |
599 | switch(base64Counter) { |
600 | case 0: |
601 | *target++=toBase64[c>>10]; |
602 | if(target<targetLimit) { |
603 | *target++=toBase64[(c>>4)&0x3f]; |
604 | if(offsets!=nullptr) { |
605 | *offsets++=sourceIndex; |
606 | *offsets++=sourceIndex++; |
607 | } |
608 | } else { |
609 | if(offsets!=nullptr) { |
610 | *offsets++=sourceIndex++; |
611 | } |
612 | cnv->charErrorBuffer[0]=toBase64[(c>>4)&0x3f]; |
613 | cnv->charErrorBufferLength=1; |
614 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
615 | } |
616 | bits=(uint8_t)((c&15)<<2); |
617 | base64Counter=1; |
618 | break; |
619 | case 1: |
620 | *target++=toBase64[bits|(c>>14)]; |
621 | if(target<targetLimit) { |
622 | *target++=toBase64[(c>>8)&0x3f]; |
623 | if(target<targetLimit) { |
624 | *target++=toBase64[(c>>2)&0x3f]; |
625 | if(offsets!=nullptr) { |
626 | *offsets++=sourceIndex; |
627 | *offsets++=sourceIndex; |
628 | *offsets++=sourceIndex++; |
629 | } |
630 | } else { |
631 | if(offsets!=nullptr) { |
632 | *offsets++=sourceIndex; |
633 | *offsets++=sourceIndex++; |
634 | } |
635 | cnv->charErrorBuffer[0]=toBase64[(c>>2)&0x3f]; |
636 | cnv->charErrorBufferLength=1; |
637 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
638 | } |
639 | } else { |
640 | if(offsets!=nullptr) { |
641 | *offsets++=sourceIndex++; |
642 | } |
643 | cnv->charErrorBuffer[0]=toBase64[(c>>8)&0x3f]; |
644 | cnv->charErrorBuffer[1]=toBase64[(c>>2)&0x3f]; |
645 | cnv->charErrorBufferLength=2; |
646 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
647 | } |
648 | bits=(uint8_t)((c&3)<<4); |
649 | base64Counter=2; |
650 | break; |
651 | case 2: |
652 | *target++=toBase64[bits|(c>>12)]; |
653 | if(target<targetLimit) { |
654 | *target++=toBase64[(c>>6)&0x3f]; |
655 | if(target<targetLimit) { |
656 | *target++=toBase64[c&0x3f]; |
657 | if(offsets!=nullptr) { |
658 | *offsets++=sourceIndex; |
659 | *offsets++=sourceIndex; |
660 | *offsets++=sourceIndex++; |
661 | } |
662 | } else { |
663 | if(offsets!=nullptr) { |
664 | *offsets++=sourceIndex; |
665 | *offsets++=sourceIndex++; |
666 | } |
667 | cnv->charErrorBuffer[0]=toBase64[c&0x3f]; |
668 | cnv->charErrorBufferLength=1; |
669 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
670 | } |
671 | } else { |
672 | if(offsets!=nullptr) { |
673 | *offsets++=sourceIndex++; |
674 | } |
675 | cnv->charErrorBuffer[0]=toBase64[(c>>6)&0x3f]; |
676 | cnv->charErrorBuffer[1]=toBase64[c&0x3f]; |
677 | cnv->charErrorBufferLength=2; |
678 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
679 | } |
680 | bits=0; |
681 | base64Counter=0; |
682 | break; |
683 | default: |
684 | /* will never occur */ |
685 | break; |
686 | } |
687 | } |
688 | } else { |
689 | /* target is full */ |
690 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
691 | break; |
692 | } |
693 | } |
694 | } |
695 | |
696 | if(pArgs->flush && source>=sourceLimit) { |
697 | /* flush remaining bits to the target */ |
698 | if(!inDirectMode) { |
699 | if (base64Counter!=0) { |
700 | if(target<targetLimit) { |
701 | *target++=toBase64[bits]; |
702 | if(offsets!=nullptr) { |
703 | *offsets++=sourceIndex-1; |
704 | } |
705 | } else { |
706 | cnv->charErrorBuffer[cnv->charErrorBufferLength++]=toBase64[bits]; |
707 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
708 | } |
709 | } |
710 | /* Add final MINUS to terminate unicodeMode */ |
711 | if(target<targetLimit) { |
712 | *target++=MINUS; |
713 | if(offsets!=nullptr) { |
714 | *offsets++=sourceIndex-1; |
715 | } |
716 | } else { |
717 | cnv->charErrorBuffer[cnv->charErrorBufferLength++]=MINUS; |
718 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
719 | } |
720 | } |
721 | /* reset the state for the next conversion */ |
722 | cnv->fromUnicodeStatus=(cnv->fromUnicodeStatus&0xf0000000)|0x1000000; /* keep version, inDirectMode=true */ |
723 | } else { |
724 | /* set the converter state back into UConverter */ |
725 | cnv->fromUnicodeStatus= |
726 | (cnv->fromUnicodeStatus&0xf0000000)| /* keep version*/ |
727 | ((uint32_t)inDirectMode<<24)|((uint32_t)base64Counter<<16)|(uint32_t)bits; |
728 | } |
729 | |
730 | /* write back the updated pointers */ |
731 | pArgs->source=source; |
732 | pArgs->target=(char *)target; |
733 | pArgs->offsets=offsets; |
734 | return; |
735 | } |
736 | |
737 | static const char * U_CALLCONV |
738 | _UTF7GetName(const UConverter *cnv) { |
739 | switch(cnv->fromUnicodeStatus>>28) { |
740 | case 1: |
741 | return "UTF-7,version=1" ; |
742 | default: |
743 | return "UTF-7" ; |
744 | } |
745 | } |
746 | U_CDECL_END |
747 | |
748 | static const UConverterImpl _UTF7Impl={ |
749 | UCNV_UTF7, |
750 | |
751 | nullptr, |
752 | nullptr, |
753 | |
754 | _UTF7Open, |
755 | nullptr, |
756 | _UTF7Reset, |
757 | |
758 | _UTF7ToUnicodeWithOffsets, |
759 | _UTF7ToUnicodeWithOffsets, |
760 | _UTF7FromUnicodeWithOffsets, |
761 | _UTF7FromUnicodeWithOffsets, |
762 | nullptr, |
763 | |
764 | nullptr, |
765 | _UTF7GetName, |
766 | nullptr, /* we don't need writeSub() because we never call a callback at fromUnicode() */ |
767 | nullptr, |
768 | ucnv_getCompleteUnicodeSet, |
769 | |
770 | nullptr, |
771 | nullptr |
772 | }; |
773 | |
774 | static const UConverterStaticData _UTF7StaticData={ |
775 | sizeof(UConverterStaticData), |
776 | "UTF-7" , |
777 | 0, /* TODO CCSID for UTF-7 */ |
778 | UCNV_IBM, UCNV_UTF7, |
779 | 1, 4, |
780 | { 0x3f, 0, 0, 0 }, 1, /* the subchar is not used */ |
781 | false, false, |
782 | 0, |
783 | 0, |
784 | { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ |
785 | }; |
786 | |
787 | const UConverterSharedData _UTF7Data= |
788 | UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_UTF7StaticData, &_UTF7Impl); |
789 | |
790 | /* IMAP mailbox name encoding ----------------------------------------------- */ |
791 | |
792 | /* |
793 | * RFC 2060: INTERNET MESSAGE ACCESS PROTOCOL - VERSION 4rev1 |
794 | * http://www.ietf.org/rfc/rfc2060.txt |
795 | * |
796 | * 5.1.3. Mailbox International Naming Convention |
797 | * |
798 | * By convention, international mailbox names are specified using a |
799 | * modified version of the UTF-7 encoding described in [UTF-7]. The |
800 | * purpose of these modifications is to correct the following problems |
801 | * with UTF-7: |
802 | * |
803 | * 1) UTF-7 uses the "+" character for shifting; this conflicts with |
804 | * the common use of "+" in mailbox names, in particular USENET |
805 | * newsgroup names. |
806 | * |
807 | * 2) UTF-7's encoding is BASE64 which uses the "/" character; this |
808 | * conflicts with the use of "/" as a popular hierarchy delimiter. |
809 | * |
810 | * 3) UTF-7 prohibits the unencoded usage of "\"; this conflicts with |
811 | * the use of "\" as a popular hierarchy delimiter. |
812 | * |
813 | * 4) UTF-7 prohibits the unencoded usage of "~"; this conflicts with |
814 | * the use of "~" in some servers as a home directory indicator. |
815 | * |
816 | * 5) UTF-7 permits multiple alternate forms to represent the same |
817 | * string; in particular, printable US-ASCII characters can be |
818 | * represented in encoded form. |
819 | * |
820 | * In modified UTF-7, printable US-ASCII characters except for "&" |
821 | * represent themselves; that is, characters with octet values 0x20-0x25 |
822 | * and 0x27-0x7e. The character "&" (0x26) is represented by the two- |
823 | * octet sequence "&-". |
824 | * |
825 | * All other characters (octet values 0x00-0x1f, 0x7f-0xff, and all |
826 | * Unicode 16-bit octets) are represented in modified BASE64, with a |
827 | * further modification from [UTF-7] that "," is used instead of "/". |
828 | * Modified BASE64 MUST NOT be used to represent any printing US-ASCII |
829 | * character which can represent itself. |
830 | * |
831 | * "&" is used to shift to modified BASE64 and "-" to shift back to US- |
832 | * ASCII. All names start in US-ASCII, and MUST end in US-ASCII (that |
833 | * is, a name that ends with a Unicode 16-bit octet MUST end with a "- |
834 | * "). |
835 | * |
836 | * For example, here is a mailbox name which mixes English, Japanese, |
837 | * and Chinese text: ~peter/mail/&ZeVnLIqe-/&U,BTFw- |
838 | */ |
839 | |
840 | /* |
841 | * Tests for US-ASCII characters belonging to character classes |
842 | * defined in UTF-7. |
843 | * |
844 | * Set D (directly encoded characters) consists of the following |
845 | * characters: the upper and lower case letters A through Z |
846 | * and a through z, the 10 digits 0-9, and the following nine special |
847 | * characters (note that "+" and "=" are omitted): |
848 | * '(),-./:? |
849 | * |
850 | * Set O (optional direct characters) consists of the following |
851 | * characters (note that "\" and "~" are omitted): |
852 | * !"#$%&*;<=>@[]^_`{|} |
853 | * |
854 | * According to the rules in RFC 2152, the byte values for the following |
855 | * US-ASCII characters are not used in UTF-7 and are therefore illegal: |
856 | * - all C0 control codes except for CR LF TAB |
857 | * - BACKSLASH |
858 | * - TILDE |
859 | * - DEL |
860 | * - all codes beyond US-ASCII, i.e. all >127 |
861 | */ |
862 | |
863 | /* uses '&' not '+' to start a base64 sequence */ |
864 | #define AMPERSAND 0x26 |
865 | #define COMMA 0x2c |
866 | #define SLASH 0x2f |
867 | |
868 | /* legal byte values: all US-ASCII graphic characters 0x20..0x7e */ |
869 | #define isLegalIMAP(c) (0x20<=(c) && (c)<=0x7e) |
870 | |
871 | /* direct-encode all of printable ASCII 0x20..0x7e except '&' 0x26 */ |
872 | #define inSetDIMAP(c) (isLegalIMAP(c) && c!=AMPERSAND) |
873 | |
874 | #define TO_BASE64_IMAP(n) ((n)<63 ? toBase64[n] : COMMA) |
875 | #define FROM_BASE64_IMAP(c) ((c)==COMMA ? 63 : (c)==SLASH ? -1 : fromBase64[c]) |
876 | |
877 | /* |
878 | * converter status values: |
879 | * |
880 | * toUnicodeStatus: |
881 | * 24 inDirectMode (boolean) |
882 | * 23..16 base64Counter (-1..7) |
883 | * 15..0 bits (up to 14 bits incoming base64) |
884 | * |
885 | * fromUnicodeStatus: |
886 | * 24 inDirectMode (boolean) |
887 | * 23..16 base64Counter (0..2) |
888 | * 7..0 bits (6 bits outgoing base64) |
889 | * |
890 | * ignore bits 31..25 |
891 | */ |
892 | |
893 | U_CDECL_BEGIN |
894 | static void U_CALLCONV |
895 | _IMAPToUnicodeWithOffsets(UConverterToUnicodeArgs *pArgs, |
896 | UErrorCode *pErrorCode) { |
897 | UConverter *cnv; |
898 | const uint8_t *source, *sourceLimit; |
899 | char16_t *target; |
900 | const char16_t *targetLimit; |
901 | int32_t *offsets; |
902 | |
903 | uint8_t *bytes; |
904 | uint8_t byteIndex; |
905 | |
906 | int32_t length, targetCapacity; |
907 | |
908 | /* UTF-7 state */ |
909 | uint16_t bits; |
910 | int8_t base64Counter; |
911 | UBool inDirectMode; |
912 | |
913 | int8_t base64Value; |
914 | |
915 | int32_t sourceIndex, nextSourceIndex; |
916 | |
917 | char16_t c; |
918 | uint8_t b; |
919 | |
920 | /* set up the local pointers */ |
921 | cnv=pArgs->converter; |
922 | |
923 | source=(const uint8_t *)pArgs->source; |
924 | sourceLimit=(const uint8_t *)pArgs->sourceLimit; |
925 | target=pArgs->target; |
926 | targetLimit=pArgs->targetLimit; |
927 | offsets=pArgs->offsets; |
928 | /* get the state machine state */ |
929 | { |
930 | uint32_t status=cnv->toUnicodeStatus; |
931 | inDirectMode=(UBool)((status>>24)&1); |
932 | base64Counter=(int8_t)(status>>16); |
933 | bits=(uint16_t)status; |
934 | } |
935 | bytes=cnv->toUBytes; |
936 | byteIndex=cnv->toULength; |
937 | |
938 | /* sourceIndex=-1 if the current character began in the previous buffer */ |
939 | sourceIndex=byteIndex==0 ? 0 : -1; |
940 | nextSourceIndex=0; |
941 | |
942 | if(inDirectMode) { |
943 | directMode: |
944 | /* |
945 | * In Direct Mode, US-ASCII characters are encoded directly, i.e., |
946 | * with their US-ASCII byte values. |
947 | * An ampersand starts Unicode (or "escape") Mode. |
948 | * |
949 | * In Direct Mode, only the sourceIndex is used. |
950 | */ |
951 | byteIndex=0; |
952 | length=(int32_t)(sourceLimit-source); |
953 | targetCapacity=(int32_t)(targetLimit-target); |
954 | if(length>targetCapacity) { |
955 | length=targetCapacity; |
956 | } |
957 | while(length>0) { |
958 | b=*source++; |
959 | if(!isLegalIMAP(b)) { |
960 | /* illegal */ |
961 | bytes[0]=b; |
962 | byteIndex=1; |
963 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
964 | break; |
965 | } else if(b!=AMPERSAND) { |
966 | /* write directly encoded character */ |
967 | *target++=b; |
968 | if(offsets!=nullptr) { |
969 | *offsets++=sourceIndex++; |
970 | } |
971 | } else /* AMPERSAND */ { |
972 | /* switch to Unicode mode */ |
973 | nextSourceIndex=++sourceIndex; |
974 | inDirectMode=false; |
975 | byteIndex=0; |
976 | bits=0; |
977 | base64Counter=-1; |
978 | goto unicodeMode; |
979 | } |
980 | --length; |
981 | } |
982 | if(source<sourceLimit && target>=targetLimit) { |
983 | /* target is full */ |
984 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
985 | } |
986 | } else { |
987 | unicodeMode: |
988 | /* |
989 | * In Unicode (or "escape") Mode, UTF-16BE is base64-encoded. |
990 | * The base64 sequence ends with any character that is not in the base64 alphabet. |
991 | * A terminating minus sign is consumed. |
992 | * US-ASCII must not be base64-ed. |
993 | * |
994 | * In Unicode Mode, the sourceIndex has the index to the start of the current |
995 | * base64 bytes, while nextSourceIndex is precisely parallel to source, |
996 | * keeping the index to the following byte. |
997 | * Note that in 2 out of 3 cases, UChars overlap within a base64 byte. |
998 | */ |
999 | while(source<sourceLimit) { |
1000 | if(target<targetLimit) { |
1001 | bytes[byteIndex++]=b=*source++; |
1002 | ++nextSourceIndex; |
1003 | if(b>0x7e) { |
1004 | /* illegal - test other illegal US-ASCII values by base64Value==-3 */ |
1005 | inDirectMode=true; |
1006 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1007 | break; |
1008 | } else if((base64Value=FROM_BASE64_IMAP(b))>=0) { |
1009 | /* collect base64 bytes into UChars */ |
1010 | switch(base64Counter) { |
1011 | case -1: /* -1 is immediately after the & */ |
1012 | case 0: |
1013 | bits=base64Value; |
1014 | base64Counter=1; |
1015 | break; |
1016 | case 1: |
1017 | case 3: |
1018 | case 4: |
1019 | case 6: |
1020 | bits=(uint16_t)((bits<<6)|base64Value); |
1021 | ++base64Counter; |
1022 | break; |
1023 | case 2: |
1024 | c=(char16_t)((bits<<4)|(base64Value>>2)); |
1025 | if(isLegalIMAP(c)) { |
1026 | /* illegal */ |
1027 | inDirectMode=true; |
1028 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1029 | goto endloop; |
1030 | } |
1031 | *target++=c; |
1032 | if(offsets!=nullptr) { |
1033 | *offsets++=sourceIndex; |
1034 | sourceIndex=nextSourceIndex-1; |
1035 | } |
1036 | bytes[0]=b; /* keep this byte in case an error occurs */ |
1037 | byteIndex=1; |
1038 | bits=(uint16_t)(base64Value&3); |
1039 | base64Counter=3; |
1040 | break; |
1041 | case 5: |
1042 | c=(char16_t)((bits<<2)|(base64Value>>4)); |
1043 | if(isLegalIMAP(c)) { |
1044 | /* illegal */ |
1045 | inDirectMode=true; |
1046 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1047 | goto endloop; |
1048 | } |
1049 | *target++=c; |
1050 | if(offsets!=nullptr) { |
1051 | *offsets++=sourceIndex; |
1052 | sourceIndex=nextSourceIndex-1; |
1053 | } |
1054 | bytes[0]=b; /* keep this byte in case an error occurs */ |
1055 | byteIndex=1; |
1056 | bits=(uint16_t)(base64Value&15); |
1057 | base64Counter=6; |
1058 | break; |
1059 | case 7: |
1060 | c=(char16_t)((bits<<6)|base64Value); |
1061 | if(isLegalIMAP(c)) { |
1062 | /* illegal */ |
1063 | inDirectMode=true; |
1064 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1065 | goto endloop; |
1066 | } |
1067 | *target++=c; |
1068 | if(offsets!=nullptr) { |
1069 | *offsets++=sourceIndex; |
1070 | sourceIndex=nextSourceIndex; |
1071 | } |
1072 | byteIndex=0; |
1073 | bits=0; |
1074 | base64Counter=0; |
1075 | break; |
1076 | default: |
1077 | /* will never occur */ |
1078 | break; |
1079 | } |
1080 | } else if(base64Value==-2) { |
1081 | /* minus sign terminates the base64 sequence */ |
1082 | inDirectMode=true; |
1083 | if(base64Counter==-1) { |
1084 | /* &- i.e. a minus immediately following an ampersand */ |
1085 | *target++=AMPERSAND; |
1086 | if(offsets!=nullptr) { |
1087 | *offsets++=sourceIndex-1; |
1088 | } |
1089 | } else { |
1090 | /* absorb the minus and leave the Unicode Mode */ |
1091 | if(bits!=0 || (base64Counter!=0 && base64Counter!=3 && base64Counter!=6)) { |
1092 | /* bits are illegally left over, a char16_t is incomplete */ |
1093 | /* base64Counter other than 0, 3, 6 means non-minimal zero-padding, also illegal */ |
1094 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1095 | break; |
1096 | } |
1097 | } |
1098 | sourceIndex=nextSourceIndex; |
1099 | goto directMode; |
1100 | } else { |
1101 | if(base64Counter==-1) { |
1102 | /* illegal: & immediately followed by something other than base64 or minus sign */ |
1103 | /* include the ampersand in the reported sequence */ |
1104 | --sourceIndex; |
1105 | bytes[0]=AMPERSAND; |
1106 | bytes[1]=b; |
1107 | byteIndex=2; |
1108 | } |
1109 | /* base64Value==-1 for characters that are illegal only in Unicode mode */ |
1110 | /* base64Value==-3 for illegal characters */ |
1111 | /* illegal */ |
1112 | inDirectMode=true; |
1113 | *pErrorCode=U_ILLEGAL_CHAR_FOUND; |
1114 | break; |
1115 | } |
1116 | } else { |
1117 | /* target is full */ |
1118 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1119 | break; |
1120 | } |
1121 | } |
1122 | } |
1123 | endloop: |
1124 | |
1125 | /* |
1126 | * the end of the input stream and detection of truncated input |
1127 | * are handled by the framework, but here we must check if we are in Unicode |
1128 | * mode and byteIndex==0 because we must end in direct mode |
1129 | * |
1130 | * conditions: |
1131 | * successful |
1132 | * in Unicode mode and byteIndex==0 |
1133 | * end of input and no truncated input |
1134 | */ |
1135 | if( U_SUCCESS(*pErrorCode) && |
1136 | !inDirectMode && byteIndex==0 && |
1137 | pArgs->flush && source>=sourceLimit |
1138 | ) { |
1139 | if(base64Counter==-1) { |
1140 | /* & at the very end of the input */ |
1141 | /* make the ampersand the reported sequence */ |
1142 | bytes[0]=AMPERSAND; |
1143 | byteIndex=1; |
1144 | } |
1145 | /* else if(base64Counter!=-1) byteIndex remains 0 because there is no particular byte sequence */ |
1146 | |
1147 | inDirectMode=true; /* avoid looping */ |
1148 | *pErrorCode=U_TRUNCATED_CHAR_FOUND; |
1149 | } |
1150 | |
1151 | /* set the converter state back into UConverter */ |
1152 | cnv->toUnicodeStatus=((uint32_t)inDirectMode<<24)|((uint32_t)((uint8_t)base64Counter)<<16)|(uint32_t)bits; |
1153 | cnv->toULength=byteIndex; |
1154 | |
1155 | /* write back the updated pointers */ |
1156 | pArgs->source=(const char *)source; |
1157 | pArgs->target=target; |
1158 | pArgs->offsets=offsets; |
1159 | return; |
1160 | } |
1161 | |
1162 | static void U_CALLCONV |
1163 | _IMAPFromUnicodeWithOffsets(UConverterFromUnicodeArgs *pArgs, |
1164 | UErrorCode *pErrorCode) { |
1165 | UConverter *cnv; |
1166 | const char16_t *source, *sourceLimit; |
1167 | uint8_t *target, *targetLimit; |
1168 | int32_t *offsets; |
1169 | |
1170 | int32_t length, targetCapacity, sourceIndex; |
1171 | char16_t c; |
1172 | uint8_t b; |
1173 | |
1174 | /* UTF-7 state */ |
1175 | uint8_t bits; |
1176 | int8_t base64Counter; |
1177 | UBool inDirectMode; |
1178 | |
1179 | /* set up the local pointers */ |
1180 | cnv=pArgs->converter; |
1181 | |
1182 | /* set up the local pointers */ |
1183 | source=pArgs->source; |
1184 | sourceLimit=pArgs->sourceLimit; |
1185 | target=(uint8_t *)pArgs->target; |
1186 | targetLimit=(uint8_t *)pArgs->targetLimit; |
1187 | offsets=pArgs->offsets; |
1188 | |
1189 | /* get the state machine state */ |
1190 | { |
1191 | uint32_t status=cnv->fromUnicodeStatus; |
1192 | inDirectMode=(UBool)((status>>24)&1); |
1193 | base64Counter=(int8_t)(status>>16); |
1194 | bits=(uint8_t)status; |
1195 | } |
1196 | |
1197 | /* UTF-7 always encodes UTF-16 code units, therefore we need only a simple sourceIndex */ |
1198 | sourceIndex=0; |
1199 | |
1200 | if(inDirectMode) { |
1201 | directMode: |
1202 | length=(int32_t)(sourceLimit-source); |
1203 | targetCapacity=(int32_t)(targetLimit-target); |
1204 | if(length>targetCapacity) { |
1205 | length=targetCapacity; |
1206 | } |
1207 | while(length>0) { |
1208 | c=*source++; |
1209 | /* encode 0x20..0x7e except '&' directly */ |
1210 | if(inSetDIMAP(c)) { |
1211 | /* encode directly */ |
1212 | *target++=(uint8_t)c; |
1213 | if(offsets!=nullptr) { |
1214 | *offsets++=sourceIndex++; |
1215 | } |
1216 | } else if(c==AMPERSAND) { |
1217 | /* output &- for & */ |
1218 | *target++=AMPERSAND; |
1219 | if(target<targetLimit) { |
1220 | *target++=MINUS; |
1221 | if(offsets!=nullptr) { |
1222 | *offsets++=sourceIndex; |
1223 | *offsets++=sourceIndex++; |
1224 | } |
1225 | /* realign length and targetCapacity */ |
1226 | goto directMode; |
1227 | } else { |
1228 | if(offsets!=nullptr) { |
1229 | *offsets++=sourceIndex++; |
1230 | } |
1231 | cnv->charErrorBuffer[0]=MINUS; |
1232 | cnv->charErrorBufferLength=1; |
1233 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1234 | break; |
1235 | } |
1236 | } else { |
1237 | /* un-read this character and switch to Unicode Mode */ |
1238 | --source; |
1239 | *target++=AMPERSAND; |
1240 | if(offsets!=nullptr) { |
1241 | *offsets++=sourceIndex; |
1242 | } |
1243 | inDirectMode=false; |
1244 | base64Counter=0; |
1245 | goto unicodeMode; |
1246 | } |
1247 | --length; |
1248 | } |
1249 | if(source<sourceLimit && target>=targetLimit) { |
1250 | /* target is full */ |
1251 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1252 | } |
1253 | } else { |
1254 | unicodeMode: |
1255 | while(source<sourceLimit) { |
1256 | if(target<targetLimit) { |
1257 | c=*source++; |
1258 | if(isLegalIMAP(c)) { |
1259 | /* encode directly */ |
1260 | inDirectMode=true; |
1261 | |
1262 | /* trick: back out this character to make this easier */ |
1263 | --source; |
1264 | |
1265 | /* terminate the base64 sequence */ |
1266 | if(base64Counter!=0) { |
1267 | /* write remaining bits for the previous character */ |
1268 | *target++=TO_BASE64_IMAP(bits); |
1269 | if(offsets!=nullptr) { |
1270 | *offsets++=sourceIndex-1; |
1271 | } |
1272 | } |
1273 | /* need to terminate with a minus */ |
1274 | if(target<targetLimit) { |
1275 | *target++=MINUS; |
1276 | if(offsets!=nullptr) { |
1277 | *offsets++=sourceIndex-1; |
1278 | } |
1279 | } else { |
1280 | cnv->charErrorBuffer[0]=MINUS; |
1281 | cnv->charErrorBufferLength=1; |
1282 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1283 | break; |
1284 | } |
1285 | goto directMode; |
1286 | } else { |
1287 | /* |
1288 | * base64 this character: |
1289 | * Output 2 or 3 base64 bytes for the remaining bits of the previous character |
1290 | * and the bits of this character, each implicitly in UTF-16BE. |
1291 | * |
1292 | * Here, bits is an 8-bit variable because only 6 bits need to be kept from one |
1293 | * character to the next. The actual 2 or 4 bits are shifted to the left edge |
1294 | * of the 6-bits field 5..0 to make the termination of the base64 sequence easier. |
1295 | */ |
1296 | switch(base64Counter) { |
1297 | case 0: |
1298 | b=(uint8_t)(c>>10); |
1299 | *target++=TO_BASE64_IMAP(b); |
1300 | if(target<targetLimit) { |
1301 | b=(uint8_t)((c>>4)&0x3f); |
1302 | *target++=TO_BASE64_IMAP(b); |
1303 | if(offsets!=nullptr) { |
1304 | *offsets++=sourceIndex; |
1305 | *offsets++=sourceIndex++; |
1306 | } |
1307 | } else { |
1308 | if(offsets!=nullptr) { |
1309 | *offsets++=sourceIndex++; |
1310 | } |
1311 | b=(uint8_t)((c>>4)&0x3f); |
1312 | cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); |
1313 | cnv->charErrorBufferLength=1; |
1314 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1315 | } |
1316 | bits=(uint8_t)((c&15)<<2); |
1317 | base64Counter=1; |
1318 | break; |
1319 | case 1: |
1320 | b=(uint8_t)(bits|(c>>14)); |
1321 | *target++=TO_BASE64_IMAP(b); |
1322 | if(target<targetLimit) { |
1323 | b=(uint8_t)((c>>8)&0x3f); |
1324 | *target++=TO_BASE64_IMAP(b); |
1325 | if(target<targetLimit) { |
1326 | b=(uint8_t)((c>>2)&0x3f); |
1327 | *target++=TO_BASE64_IMAP(b); |
1328 | if(offsets!=nullptr) { |
1329 | *offsets++=sourceIndex; |
1330 | *offsets++=sourceIndex; |
1331 | *offsets++=sourceIndex++; |
1332 | } |
1333 | } else { |
1334 | if(offsets!=nullptr) { |
1335 | *offsets++=sourceIndex; |
1336 | *offsets++=sourceIndex++; |
1337 | } |
1338 | b=(uint8_t)((c>>2)&0x3f); |
1339 | cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); |
1340 | cnv->charErrorBufferLength=1; |
1341 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1342 | } |
1343 | } else { |
1344 | if(offsets!=nullptr) { |
1345 | *offsets++=sourceIndex++; |
1346 | } |
1347 | b=(uint8_t)((c>>8)&0x3f); |
1348 | cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); |
1349 | b=(uint8_t)((c>>2)&0x3f); |
1350 | cnv->charErrorBuffer[1]=TO_BASE64_IMAP(b); |
1351 | cnv->charErrorBufferLength=2; |
1352 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1353 | } |
1354 | bits=(uint8_t)((c&3)<<4); |
1355 | base64Counter=2; |
1356 | break; |
1357 | case 2: |
1358 | b=(uint8_t)(bits|(c>>12)); |
1359 | *target++=TO_BASE64_IMAP(b); |
1360 | if(target<targetLimit) { |
1361 | b=(uint8_t)((c>>6)&0x3f); |
1362 | *target++=TO_BASE64_IMAP(b); |
1363 | if(target<targetLimit) { |
1364 | b=(uint8_t)(c&0x3f); |
1365 | *target++=TO_BASE64_IMAP(b); |
1366 | if(offsets!=nullptr) { |
1367 | *offsets++=sourceIndex; |
1368 | *offsets++=sourceIndex; |
1369 | *offsets++=sourceIndex++; |
1370 | } |
1371 | } else { |
1372 | if(offsets!=nullptr) { |
1373 | *offsets++=sourceIndex; |
1374 | *offsets++=sourceIndex++; |
1375 | } |
1376 | b=(uint8_t)(c&0x3f); |
1377 | cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); |
1378 | cnv->charErrorBufferLength=1; |
1379 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1380 | } |
1381 | } else { |
1382 | if(offsets!=nullptr) { |
1383 | *offsets++=sourceIndex++; |
1384 | } |
1385 | b=(uint8_t)((c>>6)&0x3f); |
1386 | cnv->charErrorBuffer[0]=TO_BASE64_IMAP(b); |
1387 | b=(uint8_t)(c&0x3f); |
1388 | cnv->charErrorBuffer[1]=TO_BASE64_IMAP(b); |
1389 | cnv->charErrorBufferLength=2; |
1390 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1391 | } |
1392 | bits=0; |
1393 | base64Counter=0; |
1394 | break; |
1395 | default: |
1396 | /* will never occur */ |
1397 | break; |
1398 | } |
1399 | } |
1400 | } else { |
1401 | /* target is full */ |
1402 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1403 | break; |
1404 | } |
1405 | } |
1406 | } |
1407 | |
1408 | if(pArgs->flush && source>=sourceLimit) { |
1409 | /* flush remaining bits to the target */ |
1410 | if(!inDirectMode) { |
1411 | if(base64Counter!=0) { |
1412 | if(target<targetLimit) { |
1413 | *target++=TO_BASE64_IMAP(bits); |
1414 | if(offsets!=nullptr) { |
1415 | *offsets++=sourceIndex-1; |
1416 | } |
1417 | } else { |
1418 | cnv->charErrorBuffer[cnv->charErrorBufferLength++]=TO_BASE64_IMAP(bits); |
1419 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1420 | } |
1421 | } |
1422 | /* need to terminate with a minus */ |
1423 | if(target<targetLimit) { |
1424 | *target++=MINUS; |
1425 | if(offsets!=nullptr) { |
1426 | *offsets++=sourceIndex-1; |
1427 | } |
1428 | } else { |
1429 | cnv->charErrorBuffer[cnv->charErrorBufferLength++]=MINUS; |
1430 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
1431 | } |
1432 | } |
1433 | /* reset the state for the next conversion */ |
1434 | cnv->fromUnicodeStatus=(cnv->fromUnicodeStatus&0xf0000000)|0x1000000; /* keep version, inDirectMode=true */ |
1435 | } else { |
1436 | /* set the converter state back into UConverter */ |
1437 | cnv->fromUnicodeStatus= |
1438 | (cnv->fromUnicodeStatus&0xf0000000)| /* keep version*/ |
1439 | ((uint32_t)inDirectMode<<24)|((uint32_t)base64Counter<<16)|(uint32_t)bits; |
1440 | } |
1441 | |
1442 | /* write back the updated pointers */ |
1443 | pArgs->source=source; |
1444 | pArgs->target=(char *)target; |
1445 | pArgs->offsets=offsets; |
1446 | return; |
1447 | } |
1448 | U_CDECL_END |
1449 | |
1450 | static const UConverterImpl _IMAPImpl={ |
1451 | UCNV_IMAP_MAILBOX, |
1452 | |
1453 | nullptr, |
1454 | nullptr, |
1455 | |
1456 | _UTF7Open, |
1457 | nullptr, |
1458 | _UTF7Reset, |
1459 | |
1460 | _IMAPToUnicodeWithOffsets, |
1461 | _IMAPToUnicodeWithOffsets, |
1462 | _IMAPFromUnicodeWithOffsets, |
1463 | _IMAPFromUnicodeWithOffsets, |
1464 | nullptr, |
1465 | |
1466 | nullptr, |
1467 | nullptr, |
1468 | nullptr, /* we don't need writeSub() because we never call a callback at fromUnicode() */ |
1469 | nullptr, |
1470 | ucnv_getCompleteUnicodeSet, |
1471 | nullptr, |
1472 | nullptr |
1473 | }; |
1474 | |
1475 | static const UConverterStaticData _IMAPStaticData={ |
1476 | sizeof(UConverterStaticData), |
1477 | "IMAP-mailbox-name" , |
1478 | 0, /* TODO CCSID for IMAP-mailbox-name */ |
1479 | UCNV_IBM, UCNV_IMAP_MAILBOX, |
1480 | 1, 4, |
1481 | { 0x3f, 0, 0, 0 }, 1, /* the subchar is not used */ |
1482 | false, false, |
1483 | 0, |
1484 | 0, |
1485 | { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } /* reserved */ |
1486 | }; |
1487 | |
1488 | const UConverterSharedData _IMAPData= |
1489 | UCNV_IMMUTABLE_SHARED_DATA_INITIALIZER(&_IMAPStaticData, &_IMAPImpl); |
1490 | |
1491 | #endif |
1492 | |