xref: /openbsd-src/lib/libc/net/base64.c (revision 3a3fbb3f2e2521ab7c4a56b7ff7462ebd9095ec5)
1 /*	$OpenBSD: base64.c,v 1.4 2002/01/02 23:00:10 deraadt Exp $	*/
2 
3 /*
4  * Copyright (c) 1996 by Internet Software Consortium.
5  *
6  * Permission to use, copy, modify, and distribute this software for any
7  * purpose with or without fee is hereby granted, provided that the above
8  * copyright notice and this permission notice appear in all copies.
9  *
10  * THE SOFTWARE IS PROVIDED "AS IS" AND INTERNET SOFTWARE CONSORTIUM DISCLAIMS
11  * ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES
12  * OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL INTERNET SOFTWARE
13  * CONSORTIUM BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
14  * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
15  * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
16  * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
17  * 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/types.h>
46 #include <sys/param.h>
47 #include <sys/socket.h>
48 #include <netinet/in.h>
49 #include <arpa/inet.h>
50 #include <arpa/nameser.h>
51 
52 #include <ctype.h>
53 #include <resolv.h>
54 #include <stdio.h>
55 
56 #include <stdlib.h>
57 #include <string.h>
58 
59 /* XXX abort illegal in library */
60 #define Assert(Cond) if (!(Cond)) abort()
61 
62 static const char Base64[] =
63 	"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
64 static const char Pad64 = '=';
65 
66 /* (From RFC1521 and draft-ietf-dnssec-secext-03.txt)
67    The following encoding technique is taken from RFC 1521 by Borenstein
68    and Freed.  It is reproduced here in a slightly edited form for
69    convenience.
70 
71    A 65-character subset of US-ASCII is used, enabling 6 bits to be
72    represented per printable character. (The extra 65th character, "=",
73    is used to signify a special processing function.)
74 
75    The encoding process represents 24-bit groups of input bits as output
76    strings of 4 encoded characters. Proceeding from left to right, a
77    24-bit input group is formed by concatenating 3 8-bit input groups.
78    These 24 bits are then treated as 4 concatenated 6-bit groups, each
79    of which is translated into a single digit in the base64 alphabet.
80 
81    Each 6-bit group is used as an index into an array of 64 printable
82    characters. The character referenced by the index is placed in the
83    output string.
84 
85                          Table 1: The Base64 Alphabet
86 
87       Value Encoding  Value Encoding  Value Encoding  Value Encoding
88           0 A            17 R            34 i            51 z
89           1 B            18 S            35 j            52 0
90           2 C            19 T            36 k            53 1
91           3 D            20 U            37 l            54 2
92           4 E            21 V            38 m            55 3
93           5 F            22 W            39 n            56 4
94           6 G            23 X            40 o            57 5
95           7 H            24 Y            41 p            58 6
96           8 I            25 Z            42 q            59 7
97           9 J            26 a            43 r            60 8
98          10 K            27 b            44 s            61 9
99          11 L            28 c            45 t            62 +
100          12 M            29 d            46 u            63 /
101          13 N            30 e            47 v
102          14 O            31 f            48 w         (pad) =
103          15 P            32 g            49 x
104          16 Q            33 h            50 y
105 
106    Special processing is performed if fewer than 24 bits are available
107    at the end of the data being encoded.  A full encoding quantum is
108    always completed at the end of a quantity.  When fewer than 24 input
109    bits are available in an input group, zero bits are added (on the
110    right) to form an integral number of 6-bit groups.  Padding at the
111    end of the data is performed using the '=' character.
112 
113    Since all base64 input is an integral number of octets, only the
114          -------------------------------------------------
115    following cases can arise:
116 
117        (1) the final quantum of encoding input is an integral
118            multiple of 24 bits; here, the final unit of encoded
119 	   output will be an integral multiple of 4 characters
120 	   with no "=" padding,
121        (2) the final quantum of encoding input is exactly 8 bits;
122            here, the final unit of encoded output will be two
123 	   characters followed by two "=" padding characters, or
124        (3) the final quantum of encoding input is exactly 16 bits;
125            here, the final unit of encoded output will be three
126 	   characters followed by one "=" padding character.
127    */
128 
129 int
130 b64_ntop(src, srclength, target, targsize)
131 	u_char const *src;
132 	size_t srclength;
133 	char *target;
134 	size_t targsize;
135 {
136 	size_t datalength = 0;
137 	u_char input[3];
138 	u_char output[4];
139 	int i;
140 
141 	while (2 < srclength) {
142 		input[0] = *src++;
143 		input[1] = *src++;
144 		input[2] = *src++;
145 		srclength -= 3;
146 
147 		output[0] = input[0] >> 2;
148 		output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4);
149 		output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6);
150 		output[3] = input[2] & 0x3f;
151 		Assert(output[0] < 64);
152 		Assert(output[1] < 64);
153 		Assert(output[2] < 64);
154 		Assert(output[3] < 64);
155 
156 		if (datalength + 4 > targsize)
157 			return (-1);
158 		target[datalength++] = Base64[output[0]];
159 		target[datalength++] = Base64[output[1]];
160 		target[datalength++] = Base64[output[2]];
161 		target[datalength++] = Base64[output[3]];
162 	}
163 
164 	/* Now we worry about padding. */
165 	if (0 != srclength) {
166 		/* Get what's left. */
167 		input[0] = input[1] = input[2] = '\0';
168 		for (i = 0; i < srclength; i++)
169 			input[i] = *src++;
170 
171 		output[0] = input[0] >> 2;
172 		output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4);
173 		output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6);
174 		Assert(output[0] < 64);
175 		Assert(output[1] < 64);
176 		Assert(output[2] < 64);
177 
178 		if (datalength + 4 > targsize)
179 			return (-1);
180 		target[datalength++] = Base64[output[0]];
181 		target[datalength++] = Base64[output[1]];
182 		if (srclength == 1)
183 			target[datalength++] = Pad64;
184 		else
185 			target[datalength++] = Base64[output[2]];
186 		target[datalength++] = Pad64;
187 	}
188 	if (datalength >= targsize)
189 		return (-1);
190 	target[datalength] = '\0';	/* Returned value doesn't count \0. */
191 	return (datalength);
192 }
193 
194 /* skips all whitespace anywhere.
195    converts characters, four at a time, starting at (or after)
196    src from base - 64 numbers into three 8 bit bytes in the target area.
197    it returns the number of data bytes stored at the target, or -1 on error.
198  */
199 
200 int
201 b64_pton(src, target, targsize)
202 	char const *src;
203 	u_char *target;
204 	size_t targsize;
205 {
206 	int tarindex, state, ch;
207 	char *pos;
208 
209 	state = 0;
210 	tarindex = 0;
211 
212 	while ((ch = *src++) != '\0') {
213 		if (isspace(ch))	/* Skip whitespace anywhere. */
214 			continue;
215 
216 		if (ch == Pad64)
217 			break;
218 
219 		pos = strchr(Base64, ch);
220 		if (pos == 0) 		/* A non-base64 character. */
221 			return (-1);
222 
223 		switch (state) {
224 		case 0:
225 			if (target) {
226 				if (tarindex >= targsize)
227 					return (-1);
228 				target[tarindex] = (pos - Base64) << 2;
229 			}
230 			state = 1;
231 			break;
232 		case 1:
233 			if (target) {
234 				if (tarindex + 1 >= targsize)
235 					return (-1);
236 				target[tarindex]   |=  (pos - Base64) >> 4;
237 				target[tarindex+1]  = ((pos - Base64) & 0x0f)
238 							<< 4 ;
239 			}
240 			tarindex++;
241 			state = 2;
242 			break;
243 		case 2:
244 			if (target) {
245 				if (tarindex + 1 >= targsize)
246 					return (-1);
247 				target[tarindex]   |=  (pos - Base64) >> 2;
248 				target[tarindex+1]  = ((pos - Base64) & 0x03)
249 							<< 6;
250 			}
251 			tarindex++;
252 			state = 3;
253 			break;
254 		case 3:
255 			if (target) {
256 				if (tarindex >= targsize)
257 					return (-1);
258 				target[tarindex] |= (pos - Base64);
259 			}
260 			tarindex++;
261 			state = 0;
262 			break;
263 		}
264 	}
265 
266 	/*
267 	 * We are done decoding Base-64 chars.  Let's see if we ended
268 	 * on a byte boundary, and/or with erroneous trailing characters.
269 	 */
270 
271 	if (ch == Pad64) {		/* We got a pad char. */
272 		ch = *src++;		/* Skip it, get next. */
273 		switch (state) {
274 		case 0:		/* Invalid = in first position */
275 		case 1:		/* Invalid = in second position */
276 			return (-1);
277 
278 		case 2:		/* Valid, means one byte of info */
279 			/* Skip any number of spaces. */
280 			for (; ch != '\0'; ch = *src++)
281 				if (!isspace(ch))
282 					break;
283 			/* Make sure there is another trailing = sign. */
284 			if (ch != Pad64)
285 				return (-1);
286 			ch = *src++;		/* Skip the = */
287 			/* Fall through to "single trailing =" case. */
288 			/* FALLTHROUGH */
289 
290 		case 3:		/* Valid, means two bytes of info */
291 			/*
292 			 * We know this char is an =.  Is there anything but
293 			 * whitespace after it?
294 			 */
295 			for (; ch != '\0'; ch = *src++)
296 				if (!isspace(ch))
297 					return (-1);
298 
299 			/*
300 			 * Now make sure for cases 2 and 3 that the "extra"
301 			 * bits that slopped past the last full byte were
302 			 * zeros.  If we don't check them, they become a
303 			 * subliminal channel.
304 			 */
305 			if (target && target[tarindex] != 0)
306 				return (-1);
307 		}
308 	} else {
309 		/*
310 		 * We ended by seeing the end of the string.  Make sure we
311 		 * have no partial bytes lying around.
312 		 */
313 		if (state != 0)
314 			return (-1);
315 	}
316 
317 	return (tarindex);
318 }
319