xref: /openbsd-src/lib/libc/net/base64.c (revision b2ea75c1b17e1a9a339660e7ed45cd24946b230e)
1 /*	$OpenBSD: base64.c,v 1.3 1997/11/08 20:46:55 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 #define Assert(Cond) if (!(Cond)) abort()
60 
61 static const char Base64[] =
62 	"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
63 static const char Pad64 = '=';
64 
65 /* (From RFC1521 and draft-ietf-dnssec-secext-03.txt)
66    The following encoding technique is taken from RFC 1521 by Borenstein
67    and Freed.  It is reproduced here in a slightly edited form for
68    convenience.
69 
70    A 65-character subset of US-ASCII is used, enabling 6 bits to be
71    represented per printable character. (The extra 65th character, "=",
72    is used to signify a special processing function.)
73 
74    The encoding process represents 24-bit groups of input bits as output
75    strings of 4 encoded characters. Proceeding from left to right, a
76    24-bit input group is formed by concatenating 3 8-bit input groups.
77    These 24 bits are then treated as 4 concatenated 6-bit groups, each
78    of which is translated into a single digit in the base64 alphabet.
79 
80    Each 6-bit group is used as an index into an array of 64 printable
81    characters. The character referenced by the index is placed in the
82    output string.
83 
84                          Table 1: The Base64 Alphabet
85 
86       Value Encoding  Value Encoding  Value Encoding  Value Encoding
87           0 A            17 R            34 i            51 z
88           1 B            18 S            35 j            52 0
89           2 C            19 T            36 k            53 1
90           3 D            20 U            37 l            54 2
91           4 E            21 V            38 m            55 3
92           5 F            22 W            39 n            56 4
93           6 G            23 X            40 o            57 5
94           7 H            24 Y            41 p            58 6
95           8 I            25 Z            42 q            59 7
96           9 J            26 a            43 r            60 8
97          10 K            27 b            44 s            61 9
98          11 L            28 c            45 t            62 +
99          12 M            29 d            46 u            63 /
100          13 N            30 e            47 v
101          14 O            31 f            48 w         (pad) =
102          15 P            32 g            49 x
103          16 Q            33 h            50 y
104 
105    Special processing is performed if fewer than 24 bits are available
106    at the end of the data being encoded.  A full encoding quantum is
107    always completed at the end of a quantity.  When fewer than 24 input
108    bits are available in an input group, zero bits are added (on the
109    right) to form an integral number of 6-bit groups.  Padding at the
110    end of the data is performed using the '=' character.
111 
112    Since all base64 input is an integral number of octets, only the
113          -------------------------------------------------
114    following cases can arise:
115 
116        (1) the final quantum of encoding input is an integral
117            multiple of 24 bits; here, the final unit of encoded
118 	   output will be an integral multiple of 4 characters
119 	   with no "=" padding,
120        (2) the final quantum of encoding input is exactly 8 bits;
121            here, the final unit of encoded output will be two
122 	   characters followed by two "=" padding characters, or
123        (3) the final quantum of encoding input is exactly 16 bits;
124            here, the final unit of encoded output will be three
125 	   characters followed by one "=" padding character.
126    */
127 
128 int
129 b64_ntop(src, srclength, target, targsize)
130 	u_char const *src;
131 	size_t srclength;
132 	char *target;
133 	size_t targsize;
134 {
135 	size_t datalength = 0;
136 	u_char input[3];
137 	u_char output[4];
138 	int i;
139 
140 	while (2 < srclength) {
141 		input[0] = *src++;
142 		input[1] = *src++;
143 		input[2] = *src++;
144 		srclength -= 3;
145 
146 		output[0] = input[0] >> 2;
147 		output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4);
148 		output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6);
149 		output[3] = input[2] & 0x3f;
150 		Assert(output[0] < 64);
151 		Assert(output[1] < 64);
152 		Assert(output[2] < 64);
153 		Assert(output[3] < 64);
154 
155 		if (datalength + 4 > targsize)
156 			return (-1);
157 		target[datalength++] = Base64[output[0]];
158 		target[datalength++] = Base64[output[1]];
159 		target[datalength++] = Base64[output[2]];
160 		target[datalength++] = Base64[output[3]];
161 	}
162 
163 	/* Now we worry about padding. */
164 	if (0 != srclength) {
165 		/* Get what's left. */
166 		input[0] = input[1] = input[2] = '\0';
167 		for (i = 0; i < srclength; i++)
168 			input[i] = *src++;
169 
170 		output[0] = input[0] >> 2;
171 		output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4);
172 		output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6);
173 		Assert(output[0] < 64);
174 		Assert(output[1] < 64);
175 		Assert(output[2] < 64);
176 
177 		if (datalength + 4 > targsize)
178 			return (-1);
179 		target[datalength++] = Base64[output[0]];
180 		target[datalength++] = Base64[output[1]];
181 		if (srclength == 1)
182 			target[datalength++] = Pad64;
183 		else
184 			target[datalength++] = Base64[output[2]];
185 		target[datalength++] = Pad64;
186 	}
187 	if (datalength >= targsize)
188 		return (-1);
189 	target[datalength] = '\0';	/* Returned value doesn't count \0. */
190 	return (datalength);
191 }
192 
193 /* skips all whitespace anywhere.
194    converts characters, four at a time, starting at (or after)
195    src from base - 64 numbers into three 8 bit bytes in the target area.
196    it returns the number of data bytes stored at the target, or -1 on error.
197  */
198 
199 int
200 b64_pton(src, target, targsize)
201 	char const *src;
202 	u_char *target;
203 	size_t targsize;
204 {
205 	int tarindex, state, ch;
206 	char *pos;
207 
208 	state = 0;
209 	tarindex = 0;
210 
211 	while ((ch = *src++) != '\0') {
212 		if (isspace(ch))	/* Skip whitespace anywhere. */
213 			continue;
214 
215 		if (ch == Pad64)
216 			break;
217 
218 		pos = strchr(Base64, ch);
219 		if (pos == 0) 		/* A non-base64 character. */
220 			return (-1);
221 
222 		switch (state) {
223 		case 0:
224 			if (target) {
225 				if (tarindex >= targsize)
226 					return (-1);
227 				target[tarindex] = (pos - Base64) << 2;
228 			}
229 			state = 1;
230 			break;
231 		case 1:
232 			if (target) {
233 				if (tarindex + 1 >= targsize)
234 					return (-1);
235 				target[tarindex]   |=  (pos - Base64) >> 4;
236 				target[tarindex+1]  = ((pos - Base64) & 0x0f)
237 							<< 4 ;
238 			}
239 			tarindex++;
240 			state = 2;
241 			break;
242 		case 2:
243 			if (target) {
244 				if (tarindex + 1 >= targsize)
245 					return (-1);
246 				target[tarindex]   |=  (pos - Base64) >> 2;
247 				target[tarindex+1]  = ((pos - Base64) & 0x03)
248 							<< 6;
249 			}
250 			tarindex++;
251 			state = 3;
252 			break;
253 		case 3:
254 			if (target) {
255 				if (tarindex >= targsize)
256 					return (-1);
257 				target[tarindex] |= (pos - Base64);
258 			}
259 			tarindex++;
260 			state = 0;
261 			break;
262 		}
263 	}
264 
265 	/*
266 	 * We are done decoding Base-64 chars.  Let's see if we ended
267 	 * on a byte boundary, and/or with erroneous trailing characters.
268 	 */
269 
270 	if (ch == Pad64) {		/* We got a pad char. */
271 		ch = *src++;		/* Skip it, get next. */
272 		switch (state) {
273 		case 0:		/* Invalid = in first position */
274 		case 1:		/* Invalid = in second position */
275 			return (-1);
276 
277 		case 2:		/* Valid, means one byte of info */
278 			/* Skip any number of spaces. */
279 			for (; ch != '\0'; ch = *src++)
280 				if (!isspace(ch))
281 					break;
282 			/* Make sure there is another trailing = sign. */
283 			if (ch != Pad64)
284 				return (-1);
285 			ch = *src++;		/* Skip the = */
286 			/* Fall through to "single trailing =" case. */
287 			/* FALLTHROUGH */
288 
289 		case 3:		/* Valid, means two bytes of info */
290 			/*
291 			 * We know this char is an =.  Is there anything but
292 			 * whitespace after it?
293 			 */
294 			for (; ch != '\0'; ch = *src++)
295 				if (!isspace(ch))
296 					return (-1);
297 
298 			/*
299 			 * Now make sure for cases 2 and 3 that the "extra"
300 			 * bits that slopped past the last full byte were
301 			 * zeros.  If we don't check them, they become a
302 			 * subliminal channel.
303 			 */
304 			if (target && target[tarindex] != 0)
305 				return (-1);
306 		}
307 	} else {
308 		/*
309 		 * We ended by seeing the end of the string.  Make sure we
310 		 * have no partial bytes lying around.
311 		 */
312 		if (state != 0)
313 			return (-1);
314 	}
315 
316 	return (tarindex);
317 }
318