xref: /netbsd-src/lib/libc/net/base64.c (revision a536ee5124e62c9a0051a252f7833dc8f50f44c9)
1 /*	$NetBSD: base64.c,v 1.14 2012/06/25 22:32:44 abs Exp $	*/
2 
3 /*
4  * Copyright (c) 2004 by Internet Systems Consortium, Inc. ("ISC")
5  * Copyright (c) 1996-1999 by Internet Software Consortium.
6  *
7  * Permission to use, copy, modify, and distribute this software for any
8  * purpose with or without fee is hereby granted, provided that the above
9  * copyright notice and this permission notice appear in all copies.
10  *
11  * THE SOFTWARE IS PROVIDED "AS IS" AND ISC DISCLAIMS ALL WARRANTIES
12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13  * MERCHANTABILITY AND FITNESS.  IN NO EVENT SHALL ISC BE LIABLE FOR
14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT
17  * OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18  */
19 
20 /*
21  * Portions Copyright (c) 1995 by International Business Machines, Inc.
22  *
23  * International Business Machines, Inc. (hereinafter called IBM) grants
24  * permission under its copyrights to use, copy, modify, and distribute this
25  * Software with or without fee, provided that the above copyright notice and
26  * all paragraphs of this notice appear in all copies, and that the name of IBM
27  * not be used in connection with the marketing of any product incorporating
28  * the Software or modifications thereof, without specific, written prior
29  * permission.
30  *
31  * To the extent it has a right to do so, IBM grants an immunity from suit
32  * under its patents, if any, for the use, sale or manufacture of products to
33  * the extent that such products are used for performing Domain Name System
34  * dynamic updates in TCP/IP networks by means of the Software.  No immunity is
35  * granted for any product per se or for any other function of any product.
36  *
37  * THE SOFTWARE IS PROVIDED "AS IS", AND IBM DISCLAIMS ALL WARRANTIES,
38  * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
39  * PARTICULAR PURPOSE.  IN NO EVENT SHALL IBM BE LIABLE FOR ANY SPECIAL,
40  * DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER ARISING
41  * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE, EVEN
42  * IF IBM IS APPRISED OF THE POSSIBILITY OF SUCH DAMAGES.
43  */
44 
45 #include <sys/cdefs.h>
46 #if defined(LIBC_SCCS) && !defined(lint)
47 #if 0
48 static const char rcsid[] = "Id: base64.c,v 1.4 2005/04/27 04:56:34 sra Exp";
49 #else
50 __RCSID("$NetBSD: base64.c,v 1.14 2012/06/25 22:32:44 abs Exp $");
51 #endif
52 #endif /* LIBC_SCCS and not lint */
53 
54 #include "port_before.h"
55 
56 #include <sys/types.h>
57 #include <sys/param.h>
58 #include <sys/socket.h>
59 
60 #include <netinet/in.h>
61 #include <arpa/inet.h>
62 #include <arpa/nameser.h>
63 
64 #include <assert.h>
65 #include <ctype.h>
66 #include <resolv.h>
67 #include <stdio.h>
68 #include <stdlib.h>
69 #include <string.h>
70 
71 #include "port_after.h"
72 
73 #define Assert(Cond) if (!(Cond)) abort()
74 
75 static const char Base64[] =
76 	"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
77 static const char Pad64 = '=';
78 
79 /* (From RFC1521 and draft-ietf-dnssec-secext-03.txt)
80    The following encoding technique is taken from RFC1521 by Borenstein
81    and Freed.  It is reproduced here in a slightly edited form for
82    convenience.
83 
84    A 65-character subset of US-ASCII is used, enabling 6 bits to be
85    represented per printable character. (The extra 65th character, "=",
86    is used to signify a special processing function.)
87 
88    The encoding process represents 24-bit groups of input bits as output
89    strings of 4 encoded characters. Proceeding from left to right, a
90    24-bit input group is formed by concatenating 3 8-bit input groups.
91    These 24 bits are then treated as 4 concatenated 6-bit groups, each
92    of which is translated into a single digit in the base64 alphabet.
93 
94    Each 6-bit group is used as an index into an array of 64 printable
95    characters. The character referenced by the index is placed in the
96    output string.
97 
98                          Table 1: The Base64 Alphabet
99 
100       Value Encoding  Value Encoding  Value Encoding  Value Encoding
101           0 A            17 R            34 i            51 z
102           1 B            18 S            35 j            52 0
103           2 C            19 T            36 k            53 1
104           3 D            20 U            37 l            54 2
105           4 E            21 V            38 m            55 3
106           5 F            22 W            39 n            56 4
107           6 G            23 X            40 o            57 5
108           7 H            24 Y            41 p            58 6
109           8 I            25 Z            42 q            59 7
110           9 J            26 a            43 r            60 8
111          10 K            27 b            44 s            61 9
112          11 L            28 c            45 t            62 +
113          12 M            29 d            46 u            63 /
114          13 N            30 e            47 v
115          14 O            31 f            48 w         (pad) =
116          15 P            32 g            49 x
117          16 Q            33 h            50 y
118 
119    Special processing is performed if fewer than 24 bits are available
120    at the end of the data being encoded.  A full encoding quantum is
121    always completed at the end of a quantity.  When fewer than 24 input
122    bits are available in an input group, zero bits are added (on the
123    right) to form an integral number of 6-bit groups.  Padding at the
124    end of the data is performed using the '=' character.
125 
126    Since all base64 input is an integral number of octets, only the
127          -------------------------------------------------
128    following cases can arise:
129 
130        (1) the final quantum of encoding input is an integral
131            multiple of 24 bits; here, the final unit of encoded
132 	   output will be an integral multiple of 4 characters
133 	   with no "=" padding,
134        (2) the final quantum of encoding input is exactly 8 bits;
135            here, the final unit of encoded output will be two
136 	   characters followed by two "=" padding characters, or
137        (3) the final quantum of encoding input is exactly 16 bits;
138            here, the final unit of encoded output will be three
139 	   characters followed by one "=" padding character.
140    */
141 
142 int
143 b64_ntop(u_char const *src, size_t srclength, char *target, size_t targsize) {
144 	size_t datalength = 0;
145 	u_char input[3];
146 	u_char output[4];
147 	size_t i;
148 
149 	_DIAGASSERT(src != NULL);
150 	_DIAGASSERT(target != NULL);
151 
152 	while (2U < srclength) {
153 		input[0] = *src++;
154 		input[1] = *src++;
155 		input[2] = *src++;
156 		srclength -= 3;
157 
158 		output[0] = (u_int32_t)input[0] >> 2;
159 		output[1] = ((u_int32_t)(input[0] & 0x03) << 4) +
160 		    ((u_int32_t)input[1] >> 4);
161 		output[2] = ((u_int32_t)(input[1] & 0x0f) << 2) +
162 		    ((u_int32_t)input[2] >> 6);
163 		output[3] = input[2] & 0x3f;
164 		Assert(output[0] < 64);
165 		Assert(output[1] < 64);
166 		Assert(output[2] < 64);
167 		Assert(output[3] < 64);
168 
169 		if (datalength + 4 > targsize)
170 			return (-1);
171 		target[datalength++] = Base64[output[0]];
172 		target[datalength++] = Base64[output[1]];
173 		target[datalength++] = Base64[output[2]];
174 		target[datalength++] = Base64[output[3]];
175 	}
176 
177 	/* Now we worry about padding. */
178 	if (0U != srclength) {
179 		/* Get what's left. */
180 		input[0] = input[1] = input[2] = '\0';
181 		for (i = 0; i < srclength; i++)
182 			input[i] = *src++;
183 
184 		output[0] = (u_int32_t)input[0] >> 2;
185 		output[1] = ((u_int32_t)(input[0] & 0x03) << 4) +
186 		    ((u_int32_t)input[1] >> 4);
187 		output[2] = ((u_int32_t)(input[1] & 0x0f) << 2) +
188 		    ((u_int32_t)input[2] >> 6);
189 		Assert(output[0] < 64);
190 		Assert(output[1] < 64);
191 		Assert(output[2] < 64);
192 
193 		if (datalength + 4 > targsize)
194 			return (-1);
195 		target[datalength++] = Base64[output[0]];
196 		target[datalength++] = Base64[output[1]];
197 		if (srclength == 1U)
198 			target[datalength++] = Pad64;
199 		else
200 			target[datalength++] = Base64[output[2]];
201 		target[datalength++] = Pad64;
202 	}
203 	if (datalength >= targsize)
204 		return (-1);
205 	target[datalength] = '\0';	/*%< Returned value doesn't count \\0. */
206 	_DIAGASSERT(__type_fit(int, datalength));
207 	return (int)datalength;
208 }
209 
210 /* skips all whitespace anywhere.
211    converts characters, four at a time, starting at (or after)
212    src from base - 64 numbers into three 8 bit bytes in the target area.
213    it returns the number of data bytes stored at the target, or -1 on error.
214  */
215 
216 int
217 b64_pton(char const *src, u_char *target, size_t targsize)
218 {
219 	size_t tarindex;
220 	int state, ch;
221 	char *pos;
222 
223 	_DIAGASSERT(src != NULL);
224 	_DIAGASSERT(target != NULL);
225 
226 	state = 0;
227 	tarindex = 0;
228 
229 	while ((ch = (u_char) *src++) != '\0') {
230 		if (isspace(ch))	/*%< Skip whitespace anywhere. */
231 			continue;
232 
233 		if (ch == Pad64)
234 			break;
235 
236 		pos = strchr(Base64, ch);
237 		if (pos == 0) 		/*%< A non-base64 character. */
238 			return (-1);
239 
240 		switch (state) {
241 		case 0:
242 			if (target) {
243 				if ((size_t)tarindex >= targsize)
244 					return (-1);
245 				target[tarindex] =
246 				    (unsigned char)(pos - Base64) << 2;
247 			}
248 			state = 1;
249 			break;
250 		case 1:
251 			if (target) {
252 				if ((size_t)tarindex + 1 >= targsize)
253 					return (-1);
254 				target[tarindex] |=
255 				    (u_int32_t)(pos - Base64) >> 4;
256 				target[tarindex+1]  =
257 				    (unsigned char)
258 				    (((pos - Base64) & 0x0f) << 4);
259 			}
260 			tarindex++;
261 			state = 2;
262 			break;
263 		case 2:
264 			if (target) {
265 				if ((size_t)tarindex + 1 >= targsize)
266 					return (-1);
267 				target[tarindex] |=
268 					(u_int32_t)(pos - Base64) >> 2;
269 				target[tarindex+1] =
270 				    (unsigned char)
271 				    (((pos - Base64) & 0x03) << 6);
272 			}
273 			tarindex++;
274 			state = 3;
275 			break;
276 		case 3:
277 			if (target) {
278 				if ((size_t)tarindex >= targsize)
279 					return (-1);
280 				target[tarindex] |=
281 				    (unsigned char)(pos - Base64);
282 			}
283 			tarindex++;
284 			state = 0;
285 			break;
286 		default:
287 			abort();
288 		}
289 	}
290 
291 	/*
292 	 * We are done decoding Base-64 chars.  Let's see if we ended
293 	 * on a byte boundary, and/or with erroneous trailing characters.
294 	 */
295 
296 	if (ch == Pad64) {		/*%< We got a pad char. */
297 		ch = *src++;		/*%< Skip it, get next. */
298 		switch (state) {
299 		case 0:		/*%< Invalid = in first position */
300 		case 1:		/*%< Invalid = in second position */
301 			return (-1);
302 
303 		case 2:		/*%< Valid, means one byte of info */
304 			/* Skip any number of spaces. */
305 			for (; ch != '\0'; ch = (u_char) *src++)
306 				if (!isspace(ch))
307 					break;
308 			/* Make sure there is another trailing = sign. */
309 			if (ch != Pad64)
310 				return (-1);
311 			ch = *src++;		/*%< Skip the = */
312 			/* Fall through to "single trailing =" case. */
313 			/* FALLTHROUGH */
314 
315 		case 3:		/*%< Valid, means two bytes of info */
316 			/*
317 			 * We know this char is an =.  Is there anything but
318 			 * whitespace after it?
319 			 */
320 			for (; ch != '\0'; ch = (u_char) *src++)
321 				if (!isspace(ch))
322 					return (-1);
323 
324 			/*
325 			 * Now make sure for cases 2 and 3 that the "extra"
326 			 * bits that slopped past the last full byte were
327 			 * zeros.  If we don't check them, they become a
328 			 * subliminal channel.
329 			 */
330 			if (target && target[tarindex] != 0)
331 				return (-1);
332 		}
333 	} else {
334 		/*
335 		 * We ended by seeing the end of the string.  Make sure we
336 		 * have no partial bytes lying around.
337 		 */
338 		if (state != 0)
339 			return (-1);
340 	}
341 
342 	_DIAGASSERT(__type_fit(int, tarindex));
343 	return (int)tarindex;
344 }
345 
346 /*! \file */
347