1 /*
2 * Copyright 2019-2022 The OpenSSL Project Authors. All Rights Reserved.
3 *
4 * Licensed under the Apache License 2.0 (the "License"). You may not use
5 * this file except in compliance with the License. You can obtain a copy
6 * in the file LICENSE in the source distribution or at
7 * https://www.openssl.org/source/license.html
8 */
9
10 #include <stddef.h>
11 #include <string.h>
12 #include <stdio.h>
13 #include <openssl/e_os2.h>
14 #include "crypto/punycode.h"
15
16 static const unsigned int base = 36;
17 static const unsigned int tmin = 1;
18 static const unsigned int tmax = 26;
19 static const unsigned int skew = 38;
20 static const unsigned int damp = 700;
21 static const unsigned int initial_bias = 72;
22 static const unsigned int initial_n = 0x80;
23 static const unsigned int maxint = 0xFFFFFFFF;
24 static const char delimiter = '-';
25
26 #define LABEL_BUF_SIZE 512
27
28 /*-
29 * Pseudocode:
30 *
31 * function adapt(delta,numpoints,firsttime):
32 * if firsttime then let delta = delta div damp
33 * else let delta = delta div 2
34 * let delta = delta + (delta div numpoints)
35 * let k = 0
36 * while delta > ((base - tmin) * tmax) div 2 do begin
37 * let delta = delta div (base - tmin)
38 * let k = k + base
39 * end
40 * return k + (((base - tmin + 1) * delta) div (delta + skew))
41 */
42
adapt(unsigned int delta,unsigned int numpoints,unsigned int firsttime)43 static int adapt(unsigned int delta, unsigned int numpoints,
44 unsigned int firsttime)
45 {
46 unsigned int k = 0;
47
48 delta = (firsttime) ? delta / damp : delta / 2;
49 delta = delta + delta / numpoints;
50
51 while (delta > ((base - tmin) * tmax) / 2) {
52 delta = delta / (base - tmin);
53 k = k + base;
54 }
55
56 return k + (((base - tmin + 1) * delta) / (delta + skew));
57 }
58
is_basic(unsigned int a)59 static ossl_inline int is_basic(unsigned int a)
60 {
61 return (a < 0x80) ? 1 : 0;
62 }
63
64 /*-
65 * code points digit-values
66 * ------------ ----------------------
67 * 41..5A (A-Z) = 0 to 25, respectively
68 * 61..7A (a-z) = 0 to 25, respectively
69 * 30..39 (0-9) = 26 to 35, respectively
70 */
digit_decoded(const unsigned char a)71 static ossl_inline int digit_decoded(const unsigned char a)
72 {
73 if (a >= 0x41 && a <= 0x5A)
74 return a - 0x41;
75
76 if (a >= 0x61 && a <= 0x7A)
77 return a - 0x61;
78
79 if (a >= 0x30 && a <= 0x39)
80 return a - 0x30 + 26;
81
82 return -1;
83 }
84
85 /*-
86 * Pseudocode:
87 *
88 * function ossl_punycode_decode
89 * let n = initial_n
90 * let i = 0
91 * let bias = initial_bias
92 * let output = an empty string indexed from 0
93 * consume all code points before the last delimiter (if there is one)
94 * and copy them to output, fail on any non-basic code point
95 * if more than zero code points were consumed then consume one more
96 * (which will be the last delimiter)
97 * while the input is not exhausted do begin
98 * let oldi = i
99 * let w = 1
100 * for k = base to infinity in steps of base do begin
101 * consume a code point, or fail if there was none to consume
102 * let digit = the code point's digit-value, fail if it has none
103 * let i = i + digit * w, fail on overflow
104 * let t = tmin if k <= bias {+ tmin}, or
105 * tmax if k >= bias + tmax, or k - bias otherwise
106 * if digit < t then break
107 * let w = w * (base - t), fail on overflow
108 * end
109 * let bias = adapt(i - oldi, length(output) + 1, test oldi is 0?)
110 * let n = n + i div (length(output) + 1), fail on overflow
111 * let i = i mod (length(output) + 1)
112 * {if n is a basic code point then fail}
113 * insert n into output at position i
114 * increment i
115 * end
116 */
117
ossl_punycode_decode(const char * pEncoded,const size_t enc_len,unsigned int * pDecoded,unsigned int * pout_length)118 int ossl_punycode_decode(const char *pEncoded, const size_t enc_len,
119 unsigned int *pDecoded, unsigned int *pout_length)
120 {
121 unsigned int n = initial_n;
122 unsigned int i = 0;
123 unsigned int bias = initial_bias;
124 size_t processed_in = 0, written_out = 0;
125 unsigned int max_out = *pout_length;
126 unsigned int basic_count = 0;
127 unsigned int loop;
128
129 for (loop = 0; loop < enc_len; loop++) {
130 if (pEncoded[loop] == delimiter)
131 basic_count = loop;
132 }
133
134 if (basic_count > 0) {
135 if (basic_count > max_out)
136 return 0;
137
138 for (loop = 0; loop < basic_count; loop++) {
139 if (is_basic(pEncoded[loop]) == 0)
140 return 0;
141
142 pDecoded[loop] = pEncoded[loop];
143 written_out++;
144 }
145 processed_in = basic_count + 1;
146 }
147
148 for (loop = processed_in; loop < enc_len;) {
149 unsigned int oldi = i;
150 unsigned int w = 1;
151 unsigned int k, t;
152 int digit;
153
154 for (k = base;; k += base) {
155 if (loop >= enc_len)
156 return 0;
157
158 digit = digit_decoded(pEncoded[loop]);
159 loop++;
160
161 if (digit < 0)
162 return 0;
163 if ((unsigned int)digit > (maxint - i) / w)
164 return 0;
165
166 i = i + digit * w;
167 t = (k <= bias) ? tmin : (k >= bias + tmax) ? tmax : k - bias;
168
169 if ((unsigned int)digit < t)
170 break;
171
172 if (w > maxint / (base - t))
173 return 0;
174 w = w * (base - t);
175 }
176
177 bias = adapt(i - oldi, written_out + 1, (oldi == 0));
178 if (i / (written_out + 1) > maxint - n)
179 return 0;
180 n = n + i / (written_out + 1);
181 i %= (written_out + 1);
182
183 if (written_out >= max_out)
184 return 0;
185
186 memmove(pDecoded + i + 1, pDecoded + i,
187 (written_out - i) * sizeof(*pDecoded));
188 pDecoded[i] = n;
189 i++;
190 written_out++;
191 }
192
193 *pout_length = written_out;
194 return 1;
195 }
196
197 /*
198 * Encode a code point using UTF-8
199 * return number of bytes on success, 0 on failure
200 * (also produces U+FFFD, which uses 3 bytes on failure)
201 */
codepoint2utf8(unsigned char * out,unsigned long utf)202 static int codepoint2utf8(unsigned char *out, unsigned long utf)
203 {
204 if (utf <= 0x7F) {
205 /* Plain ASCII */
206 out[0] = (unsigned char)utf;
207 out[1] = 0;
208 return 1;
209 } else if (utf <= 0x07FF) {
210 /* 2-byte unicode */
211 out[0] = (unsigned char)(((utf >> 6) & 0x1F) | 0xC0);
212 out[1] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
213 out[2] = 0;
214 return 2;
215 } else if (utf <= 0xFFFF) {
216 /* 3-byte unicode */
217 out[0] = (unsigned char)(((utf >> 12) & 0x0F) | 0xE0);
218 out[1] = (unsigned char)(((utf >> 6) & 0x3F) | 0x80);
219 out[2] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
220 out[3] = 0;
221 return 3;
222 } else if (utf <= 0x10FFFF) {
223 /* 4-byte unicode */
224 out[0] = (unsigned char)(((utf >> 18) & 0x07) | 0xF0);
225 out[1] = (unsigned char)(((utf >> 12) & 0x3F) | 0x80);
226 out[2] = (unsigned char)(((utf >> 6) & 0x3F) | 0x80);
227 out[3] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
228 out[4] = 0;
229 return 4;
230 } else {
231 /* error - use replacement character */
232 out[0] = (unsigned char)0xEF;
233 out[1] = (unsigned char)0xBF;
234 out[2] = (unsigned char)0xBD;
235 out[3] = 0;
236 return 0;
237 }
238 }
239
240 /*-
241 * Return values:
242 * 1 - ok, *outlen contains valid buf length
243 * 0 - ok but buf was too short, *outlen contains valid buf length
244 * -1 - bad string passed
245 */
246
ossl_a2ulabel(const char * in,char * out,size_t * outlen)247 int ossl_a2ulabel(const char *in, char *out, size_t *outlen)
248 {
249 /*-
250 * Domain name has some parts consisting of ASCII chars joined with dot.
251 * If a part is shorter than 5 chars, it becomes U-label as is.
252 * If it does not start with xn--, it becomes U-label as is.
253 * Otherwise we try to decode it.
254 */
255 char *outptr = out;
256 const char *inptr = in;
257 size_t size = 0, maxsize;
258 int result = 1;
259 unsigned int i, j;
260 unsigned int buf[LABEL_BUF_SIZE]; /* It's a hostname */
261
262 if (out == NULL) {
263 result = 0;
264 maxsize = 0;
265 } else {
266 maxsize = *outlen;
267 }
268
269 #define PUSHC(c) \
270 do \
271 if (size++ < maxsize) \
272 *outptr++ = c; \
273 else \
274 result = 0; \
275 while (0)
276
277 while (1) {
278 char *tmpptr = strchr(inptr, '.');
279 size_t delta = tmpptr != NULL ? (size_t)(tmpptr - inptr) : strlen(inptr);
280
281 if (strncmp(inptr, "xn--", 4) != 0) {
282 for (i = 0; i < delta + 1; i++)
283 PUSHC(inptr[i]);
284 } else {
285 unsigned int bufsize = LABEL_BUF_SIZE;
286
287 if (ossl_punycode_decode(inptr + 4, delta - 4, buf, &bufsize) <= 0)
288 return -1;
289
290 for (i = 0; i < bufsize; i++) {
291 unsigned char seed[6];
292 size_t utfsize = codepoint2utf8(seed, buf[i]);
293
294 if (utfsize == 0)
295 return -1;
296
297 for (j = 0; j < utfsize; j++)
298 PUSHC(seed[j]);
299 }
300
301 PUSHC(tmpptr != NULL ? '.' : '\0');
302 }
303
304 if (tmpptr == NULL)
305 break;
306
307 inptr = tmpptr + 1;
308 }
309 #undef PUSHC
310
311 *outlen = size;
312 return result;
313 }
314
315 /*-
316 * a MUST be A-label
317 * u MUST be U-label
318 * Returns 0 if compared values are equal
319 * 1 if not
320 * -1 in case of errors
321 */
322
ossl_a2ucompare(const char * a,const char * u)323 int ossl_a2ucompare(const char *a, const char *u)
324 {
325 char a_ulabel[LABEL_BUF_SIZE + 1];
326 size_t a_size = sizeof(a_ulabel);
327
328 if (ossl_a2ulabel(a, a_ulabel, &a_size) <= 0)
329 return -1;
330
331 return strcmp(a_ulabel, u) != 0;
332 }
333