2 * Copyright 2019-2020 The OpenSSL Project Authors. All Rights Reserved.
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
13 #include <openssl/e_os2.h>
14 #include "crypto/punycode.h"
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 = '-';
26 #define LABEL_BUF_SIZE 512
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)
36 * while delta > ((base - tmin) * tmax) div 2 do begin
37 * let delta = delta div (base - tmin)
40 * return k + (((base - tmin + 1) * delta) div (delta + skew))
43 static int adapt(unsigned int delta, unsigned int numpoints,
44 unsigned int firsttime)
48 delta = (firsttime) ? delta / damp : delta / 2;
49 delta = delta + delta / numpoints;
51 while (delta > ((base - tmin) * tmax) / 2) {
52 delta = delta / (base - tmin);
56 return k + (((base - tmin + 1) * delta) / (delta + skew));
59 static ossl_inline int is_basic(unsigned int a)
61 return (a < 0x80) ? 1 : 0;
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
71 static ossl_inline int digit_decoded(const unsigned char a)
73 if (a >= 0x41 && a <= 0x5A)
76 if (a >= 0x61 && a <= 0x7A)
79 if (a >= 0x30 && a <= 0x39)
88 * function ossl_punycode_decode
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
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
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
118 int ossl_punycode_decode(const char *pEncoded, const size_t enc_len,
119 unsigned int *pDecoded, unsigned int *pout_length)
121 unsigned int n = initial_n;
123 unsigned int bias = initial_bias;
124 size_t processed_in = 0, written_out = 0;
125 unsigned int max_out = *pout_length;
127 unsigned int basic_count = 0;
130 for (loop = 0; loop < enc_len; loop++) {
131 if (pEncoded[loop] == delimiter)
135 if (basic_count > 0) {
136 if (basic_count > max_out)
139 for (loop = 0; loop < basic_count; loop++) {
140 if (is_basic(pEncoded[loop]) == 0)
143 pDecoded[loop] = pEncoded[loop];
146 processed_in = basic_count + 1;
149 for (loop = processed_in; loop < enc_len;) {
150 unsigned int oldi = i;
155 for (k = base;; k += base) {
159 digit = digit_decoded(pEncoded[loop]);
164 if ((unsigned int)digit > (maxint - i) / w)
168 t = (k <= bias) ? tmin : (k >= bias + tmax) ? tmax : k - bias;
170 if ((unsigned int)digit < t)
173 if (w > maxint / (base - t))
178 bias = adapt(i - oldi, written_out + 1, (oldi == 0));
179 if (i / (written_out + 1) > maxint - n)
181 n = n + i / (written_out + 1);
182 i %= (written_out + 1);
184 if (written_out > max_out)
187 memmove(pDecoded + i + 1, pDecoded + i,
188 (written_out - i) * sizeof *pDecoded);
194 *pout_length = written_out;
199 * Encode a code point using UTF-8
200 * return number of bytes on success, 0 on failure
201 * (also produces U+FFFD, which uses 3 bytes on failure)
203 static int codepoint2utf8(unsigned char *out, unsigned long utf)
207 out[0] = (unsigned char)utf;
210 } else if (utf <= 0x07FF) {
212 out[0] = (unsigned char)(((utf >> 6) & 0x1F) | 0xC0);
213 out[1] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
216 } else if (utf <= 0xFFFF) {
218 out[0] = (unsigned char)(((utf >> 12) & 0x0F) | 0xE0);
219 out[1] = (unsigned char)(((utf >> 6) & 0x3F) | 0x80);
220 out[2] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
223 } else if (utf <= 0x10FFFF) {
225 out[0] = (unsigned char)(((utf >> 18) & 0x07) | 0xF0);
226 out[1] = (unsigned char)(((utf >> 12) & 0x3F) | 0x80);
227 out[2] = (unsigned char)(((utf >> 6) & 0x3F) | 0x80);
228 out[3] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
232 /* error - use replacement character */
233 out[0] = (unsigned char)0xEF;
234 out[1] = (unsigned char)0xBF;
235 out[2] = (unsigned char)0xBD;
243 * 1 - ok, *outlen contains valid buf length
244 * 0 - ok but buf was too short, *outlen contains valid buf length
245 * -1 - bad string passed
248 int ossl_a2ulabel(const char *in, char *out, size_t *outlen)
251 * Domain name has some parts consisting of ASCII chars joined with dot.
252 * If a part is shorter than 5 chars, it becomes U-label as is.
253 * If it does not start with xn--, it becomes U-label as is.
254 * Otherwise we try to decode it.
257 const char *inptr = in;
261 unsigned int buf[LABEL_BUF_SIZE]; /* It's a hostname */
266 char *tmpptr = strchr(inptr, '.');
267 size_t delta = (tmpptr) ? (size_t)(tmpptr - inptr) : strlen(inptr);
269 if (strncmp(inptr, "xn--", 4) != 0) {
272 if (size >= *outlen - 1)
276 memcpy(outptr, inptr, delta + 1);
280 unsigned int bufsize = LABEL_BUF_SIZE;
283 if (ossl_punycode_decode(inptr + 4, delta - 4, buf, &bufsize) <= 0)
286 for (i = 0; i < bufsize; i++) {
287 unsigned char seed[6];
288 size_t utfsize = codepoint2utf8(seed, buf[i]);
293 if (size >= *outlen - 1)
297 memcpy(outptr, seed, utfsize);
302 if (tmpptr != NULL) {
306 if (size >= *outlen - 1)
323 * Returns 0 if compared values are equal
325 * -1 in case of errors
328 int ossl_a2ucompare(const char *a, const char *u)
330 char a_ulabel[LABEL_BUF_SIZE];
331 size_t a_size = sizeof(a_ulabel);
333 if (ossl_a2ulabel(a, a_ulabel, &a_size) <= 0) {
337 return (strcmp(a_ulabel, u) == 0) ? 0 : 1;