1 /* Copyright libuv contributors. All rights reserved.
2 *
3 * Permission to use, copy, modify, and/or distribute this software for any
4 * purpose with or without fee is hereby granted, provided that the above
5 * copyright notice and this permission notice appear in all copies.
6 *
7 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
8 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
9 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
10 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
11 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
12 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
13 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
14 */
15
16 /* Derived from https://github.com/bnoordhuis/punycode
17 * but updated to support IDNA 2008.
18 */
19
20 #include "uv.h"
21 #include "uv-common.h"
22 #include "idna.h"
23 #include <assert.h>
24 #include <string.h>
25 #include <limits.h> /* UINT_MAX */
26
27
uv__wtf8_decode1(const char ** input)28 static int32_t uv__wtf8_decode1(const char** input) {
29 uint32_t code_point;
30 uint8_t b1;
31 uint8_t b2;
32 uint8_t b3;
33 uint8_t b4;
34
35 b1 = **input;
36 if (b1 <= 0x7F)
37 return b1; /* ASCII code point */
38 if (b1 < 0xC2)
39 return -1; /* invalid: continuation byte */
40 code_point = b1;
41
42 b2 = *++*input;
43 if ((b2 & 0xC0) != 0x80)
44 return -1; /* invalid: not a continuation byte */
45 code_point = (code_point << 6) | (b2 & 0x3F);
46 if (b1 <= 0xDF)
47 return 0x7FF & code_point; /* two-byte character */
48
49 b3 = *++*input;
50 if ((b3 & 0xC0) != 0x80)
51 return -1; /* invalid: not a continuation byte */
52 code_point = (code_point << 6) | (b3 & 0x3F);
53 if (b1 <= 0xEF)
54 return 0xFFFF & code_point; /* three-byte character */
55
56 b4 = *++*input;
57 if ((b4 & 0xC0) != 0x80)
58 return -1; /* invalid: not a continuation byte */
59 code_point = (code_point << 6) | (b4 & 0x3F);
60 if (b1 <= 0xF4) {
61 code_point &= 0x1FFFFF;
62 if (code_point <= 0x10FFFF)
63 return code_point; /* four-byte character */
64 }
65
66 /* code point too large */
67 return -1;
68 }
69
70
uv__utf8_decode1_slow(const char ** p,const char * pe,unsigned a)71 static unsigned uv__utf8_decode1_slow(const char** p,
72 const char* pe,
73 unsigned a) {
74 unsigned b;
75 unsigned c;
76 unsigned d;
77 unsigned min;
78
79 if (a > 0xF7)
80 return -1;
81
82 switch (pe - *p) {
83 default:
84 if (a > 0xEF) {
85 min = 0x10000;
86 a = a & 7;
87 b = (unsigned char) *(*p)++;
88 c = (unsigned char) *(*p)++;
89 d = (unsigned char) *(*p)++;
90 break;
91 }
92 /* Fall through. */
93 case 2:
94 if (a > 0xDF) {
95 min = 0x800;
96 b = 0x80 | (a & 15);
97 c = (unsigned char) *(*p)++;
98 d = (unsigned char) *(*p)++;
99 a = 0;
100 break;
101 }
102 /* Fall through. */
103 case 1:
104 if (a > 0xBF) {
105 min = 0x80;
106 b = 0x80;
107 c = 0x80 | (a & 31);
108 d = (unsigned char) *(*p)++;
109 a = 0;
110 break;
111 }
112 /* Fall through. */
113 case 0:
114 return -1; /* Invalid continuation byte. */
115 }
116
117 if (0x80 != (0xC0 & (b ^ c ^ d)))
118 return -1; /* Invalid sequence. */
119
120 b &= 63;
121 c &= 63;
122 d &= 63;
123 a = (a << 18) | (b << 12) | (c << 6) | d;
124
125 if (a < min)
126 return -1; /* Overlong sequence. */
127
128 if (a > 0x10FFFF)
129 return -1; /* Four-byte sequence > U+10FFFF. */
130
131 if (a >= 0xD800 && a <= 0xDFFF)
132 return -1; /* Surrogate pair. */
133
134 return a;
135 }
136
137
uv__utf8_decode1(const char ** p,const char * pe)138 unsigned uv__utf8_decode1(const char** p, const char* pe) {
139 unsigned a;
140
141 assert(*p < pe);
142
143 a = (unsigned char) *(*p)++;
144
145 if (a < 128)
146 return a; /* ASCII, common case. */
147
148 return uv__utf8_decode1_slow(p, pe, a);
149 }
150
151
uv__idna_toascii_label(const char * s,const char * se,char ** d,char * de)152 static int uv__idna_toascii_label(const char* s, const char* se,
153 char** d, char* de) {
154 static const char alphabet[] = "abcdefghijklmnopqrstuvwxyz0123456789";
155 const char* ss;
156 unsigned c;
157 unsigned h;
158 unsigned k;
159 unsigned n;
160 unsigned m;
161 unsigned q;
162 unsigned t;
163 unsigned x;
164 unsigned y;
165 unsigned bias;
166 unsigned delta;
167 unsigned todo;
168 int first;
169
170 h = 0;
171 ss = s;
172 todo = 0;
173
174 /* Note: after this loop we've visited all UTF-8 characters and know
175 * they're legal so we no longer need to check for decode errors.
176 */
177 while (s < se) {
178 c = uv__utf8_decode1(&s, se);
179
180 if (c == UINT_MAX)
181 return UV_EINVAL;
182
183 if (c < 128)
184 h++;
185 else
186 todo++;
187 }
188
189 /* Only write "xn--" when there are non-ASCII characters. */
190 if (todo > 0) {
191 if (*d < de) *(*d)++ = 'x';
192 if (*d < de) *(*d)++ = 'n';
193 if (*d < de) *(*d)++ = '-';
194 if (*d < de) *(*d)++ = '-';
195 }
196
197 /* Write ASCII characters. */
198 x = 0;
199 s = ss;
200 while (s < se) {
201 c = uv__utf8_decode1(&s, se);
202 assert(c != UINT_MAX);
203
204 if (c > 127)
205 continue;
206
207 if (*d < de)
208 *(*d)++ = c;
209
210 if (++x == h)
211 break; /* Visited all ASCII characters. */
212 }
213
214 if (todo == 0)
215 return h;
216
217 /* Only write separator when we've written ASCII characters first. */
218 if (h > 0)
219 if (*d < de)
220 *(*d)++ = '-';
221
222 n = 128;
223 bias = 72;
224 delta = 0;
225 first = 1;
226
227 while (todo > 0) {
228 m = -1;
229 s = ss;
230
231 while (s < se) {
232 c = uv__utf8_decode1(&s, se);
233 assert(c != UINT_MAX);
234
235 if (c >= n)
236 if (c < m)
237 m = c;
238 }
239
240 x = m - n;
241 y = h + 1;
242
243 if (x > ~delta / y)
244 return UV_E2BIG; /* Overflow. */
245
246 delta += x * y;
247 n = m;
248
249 s = ss;
250 while (s < se) {
251 c = uv__utf8_decode1(&s, se);
252 assert(c != UINT_MAX);
253
254 if (c < n)
255 if (++delta == 0)
256 return UV_E2BIG; /* Overflow. */
257
258 if (c != n)
259 continue;
260
261 for (k = 36, q = delta; /* empty */; k += 36) {
262 t = 1;
263
264 if (k > bias)
265 t = k - bias;
266
267 if (t > 26)
268 t = 26;
269
270 if (q < t)
271 break;
272
273 /* TODO(bnoordhuis) Since 1 <= t <= 26 and therefore
274 * 10 <= y <= 35, we can optimize the long division
275 * into a table-based reciprocal multiplication.
276 */
277 x = q - t;
278 y = 36 - t; /* 10 <= y <= 35 since 1 <= t <= 26. */
279 q = x / y;
280 t = t + x % y; /* 1 <= t <= 35 because of y. */
281
282 if (*d < de)
283 *(*d)++ = alphabet[t];
284 }
285
286 if (*d < de)
287 *(*d)++ = alphabet[q];
288
289 delta /= 2;
290
291 if (first) {
292 delta /= 350;
293 first = 0;
294 }
295
296 /* No overflow check is needed because |delta| was just
297 * divided by 2 and |delta+delta >= delta + delta/h|.
298 */
299 h++;
300 delta += delta / h;
301
302 for (bias = 0; delta > 35 * 26 / 2; bias += 36)
303 delta /= 35;
304
305 bias += 36 * delta / (delta + 38);
306 delta = 0;
307 todo--;
308 }
309
310 delta++;
311 n++;
312 }
313
314 return 0;
315 }
316
317
uv__idna_toascii(const char * s,const char * se,char * d,char * de)318 ssize_t uv__idna_toascii(const char* s, const char* se, char* d, char* de) {
319 const char* si;
320 const char* st;
321 unsigned c;
322 char* ds;
323 int rc;
324
325 if (s == se)
326 return UV_EINVAL;
327
328 ds = d;
329
330 si = s;
331 while (si < se) {
332 st = si;
333 c = uv__utf8_decode1(&si, se);
334
335 if (c == UINT_MAX)
336 return UV_EINVAL;
337
338 if (c != '.')
339 if (c != 0x3002) /* 。 */
340 if (c != 0xFF0E) /* ． */
341 if (c != 0xFF61) /* ｡ */
342 continue;
343
344 rc = uv__idna_toascii_label(s, st, &d, de);
345
346 if (rc < 0)
347 return rc;
348
349 if (d < de)
350 *d++ = '.';
351
352 s = si;
353 }
354
355 if (s < se) {
356 rc = uv__idna_toascii_label(s, se, &d, de);
357
358 if (rc < 0)
359 return rc;
360 }
361
362 if (d >= de)
363 return UV_EINVAL;
364
365 *d++ = '\0';
366 return d - ds; /* Number of bytes written. */
367 }
368
369
uv_wtf8_length_as_utf16(const char * source_ptr)370 ssize_t uv_wtf8_length_as_utf16(const char* source_ptr) {
371 size_t w_target_len = 0;
372 int32_t code_point;
373
374 do {
375 code_point = uv__wtf8_decode1(&source_ptr);
376 if (code_point < 0)
377 return -1;
378 if (code_point > 0xFFFF)
379 w_target_len++;
380 w_target_len++;
381 } while (*source_ptr++);
382
383 return w_target_len;
384 }
385
386
uv_wtf8_to_utf16(const char * source_ptr,uint16_t * w_target,size_t w_target_len)387 void uv_wtf8_to_utf16(const char* source_ptr,
388 uint16_t* w_target,
389 size_t w_target_len) {
390 int32_t code_point;
391
392 do {
393 code_point = uv__wtf8_decode1(&source_ptr);
394 /* uv_wtf8_length_as_utf16 should have been called and checked first. */
395 assert(code_point >= 0);
396 if (code_point > 0x10000) {
397 assert(code_point < 0x10FFFF);
398 *w_target++ = (((code_point - 0x10000) >> 10) + 0xD800);
399 *w_target++ = ((code_point - 0x10000) & 0x3FF) + 0xDC00;
400 w_target_len -= 2;
401 } else {
402 *w_target++ = code_point;
403 w_target_len -= 1;
404 }
405 } while (*source_ptr++);
406
407 (void)w_target_len;
408 assert(w_target_len == 0);
409 }
410
411
uv__get_surrogate_value(const uint16_t * w_source_ptr,ssize_t w_source_len)412 static int32_t uv__get_surrogate_value(const uint16_t* w_source_ptr,
413 ssize_t w_source_len) {
414 uint16_t u;
415 uint16_t next;
416
417 u = w_source_ptr[0];
418 if (u >= 0xD800 && u <= 0xDBFF && w_source_len != 1) {
419 next = w_source_ptr[1];
420 if (next >= 0xDC00 && next <= 0xDFFF)
421 return 0x10000 + ((u - 0xD800) << 10) + (next - 0xDC00);
422 }
423 return u;
424 }
425
426
uv_utf16_length_as_wtf8(const uint16_t * w_source_ptr,ssize_t w_source_len)427 size_t uv_utf16_length_as_wtf8(const uint16_t* w_source_ptr,
428 ssize_t w_source_len) {
429 size_t target_len;
430 int32_t code_point;
431
432 target_len = 0;
433 while (w_source_len) {
434 code_point = uv__get_surrogate_value(w_source_ptr, w_source_len);
435 /* Can be invalid UTF-8 but must be valid WTF-8. */
436 assert(code_point >= 0);
437 if (w_source_len < 0 && code_point == 0)
438 break;
439 if (code_point < 0x80)
440 target_len += 1;
441 else if (code_point < 0x800)
442 target_len += 2;
443 else if (code_point < 0x10000)
444 target_len += 3;
445 else {
446 target_len += 4;
447 w_source_ptr++;
448 if (w_source_len > 0)
449 w_source_len--;
450 }
451 w_source_ptr++;
452 if (w_source_len > 0)
453 w_source_len--;
454 }
455
456 return target_len;
457 }
458
459
uv_utf16_to_wtf8(const uint16_t * w_source_ptr,ssize_t w_source_len,char ** target_ptr,size_t * target_len_ptr)460 int uv_utf16_to_wtf8(const uint16_t* w_source_ptr,
461 ssize_t w_source_len,
462 char** target_ptr,
463 size_t* target_len_ptr) {
464 size_t target_len;
465 char* target;
466 char* target_end;
467 int32_t code_point;
468
469 /* If *target_ptr is provided, then *target_len_ptr must be its length
470 * (excluding space for NUL), otherwise we will compute the target_len_ptr
471 * length and may return a new allocation in *target_ptr if target_ptr is
472 * provided. */
473 if (target_ptr == NULL || *target_ptr == NULL) {
474 target_len = uv_utf16_length_as_wtf8(w_source_ptr, w_source_len);
475 if (target_len_ptr != NULL)
476 *target_len_ptr = target_len;
477 } else {
478 target_len = *target_len_ptr;
479 }
480
481 if (target_ptr == NULL)
482 return 0;
483
484 if (*target_ptr == NULL) {
485 target = uv__malloc(target_len + 1);
486 if (target == NULL) {
487 return UV_ENOMEM;
488 }
489 *target_ptr = target;
490 } else {
491 target = *target_ptr;
492 }
493
494 target_end = target + target_len;
495
496 while (target != target_end && w_source_len) {
497 code_point = uv__get_surrogate_value(w_source_ptr, w_source_len);
498 /* Can be invalid UTF-8 but must be valid WTF-8. */
499 assert(code_point >= 0);
500 if (w_source_len < 0 && code_point == 0) {
501 w_source_len = 0;
502 break;
503 }
504 if (code_point < 0x80) {
505 *target++ = code_point;
506 } else if (code_point < 0x800) {
507 *target++ = 0xC0 | (code_point >> 6);
508 if (target == target_end)
509 break;
510 *target++ = 0x80 | (code_point & 0x3F);
511 } else if (code_point < 0x10000) {
512 *target++ = 0xE0 | (code_point >> 12);
513 if (target == target_end)
514 break;
515 *target++ = 0x80 | ((code_point >> 6) & 0x3F);
516 if (target == target_end)
517 break;
518 *target++ = 0x80 | (code_point & 0x3F);
519 } else {
520 *target++ = 0xF0 | (code_point >> 18);
521 if (target == target_end)
522 break;
523 *target++ = 0x80 | ((code_point >> 12) & 0x3F);
524 if (target == target_end)
525 break;
526 *target++ = 0x80 | ((code_point >> 6) & 0x3F);
527 if (target == target_end)
528 break;
529 *target++ = 0x80 | (code_point & 0x3F);
530 /* uv__get_surrogate_value consumed 2 input characters */
531 w_source_ptr++;
532 if (w_source_len > 0)
533 w_source_len--;
534 }
535 target_len = target - *target_ptr;
536 w_source_ptr++;
537 if (w_source_len > 0)
538 w_source_len--;
539 }
540
541 if (target != target_end && target_len_ptr != NULL)
542 /* Did not fill all of the provided buffer, so update the target_len_ptr
543 * output with the space used. */
544 *target_len_ptr = target - *target_ptr;
545
546 /* Check if input fit into target exactly. */
547 if (w_source_len < 0 && target == target_end && w_source_ptr[0] == 0)
548 w_source_len = 0;
549
550 *target++ = '\0';
551
552 /* Characters remained after filling the buffer, compute the remaining length now. */
553 if (w_source_len) {
554 if (target_len_ptr != NULL)
555 *target_len_ptr = target_len + uv_utf16_length_as_wtf8(w_source_ptr, w_source_len);
556 return UV_ENOBUFS;
557 }
558
559 return 0;
560 }
561
