1 /*
2  * Copyright (c) 1996, 1998 by Internet Software Consortium.
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND INTERNET SOFTWARE CONSORTIUM DISCLAIMS
9  * ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES
10  * OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL INTERNET SOFTWARE
11  * CONSORTIUM BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
12  * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
13  * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
14  * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
15  * SOFTWARE.
16  */
17 
18 /*
19  * Portions Copyright (c) 1995 by International Business Machines, Inc.
20  *
21  * International Business Machines, Inc. (hereinafter called IBM) grants
22  * permission under its copyrights to use, copy, modify, and distribute this
23  * Software with or without fee, provided that the above copyright notice and
24  * all paragraphs of this notice appear in all copies, and that the name of IBM
25  * not be used in connection with the marketing of any product incorporating
26  * the Software or modifications thereof, without specific, written prior
27  * permission.
28  *
29  * To the extent it has a right to do so, IBM grants an immunity from suit
30  * under its patents, if any, for the use, sale or manufacture of products to
31  * the extent that such products are used for performing Domain Name System
32  * dynamic updates in TCP/IP networks by means of the Software.  No immunity is
33  * granted for any product per se or for any other function of any product.
34  *
35  * THE SOFTWARE IS PROVIDED "AS IS", AND IBM DISCLAIMS ALL WARRANTIES,
36  * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
37  * PARTICULAR PURPOSE.  IN NO EVENT SHALL IBM BE LIABLE FOR ANY SPECIAL,
38  * DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER ARISING
39  * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE, EVEN
40  * IF IBM IS APPRISED OF THE POSSIBILITY OF SUCH DAMAGES.
41  */
42 #include <config.h>
43 
44 #include <sys/types.h>
45 #include <sys/param.h>
46 #include <sys/socket.h>
47 
48 #include <netinet/in.h>
49 #include <arpa/inet.h>
50 
51 #include <ctype.h>
52 #include <stdio.h>
53 #include <stdlib.h>
54 #include <string.h>
55 
56 #define Assert(Cond) if (!(Cond)) abort()
57 
58 static const char Base64[] =
59 	"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
60 static const char Pad64 = '=';
61 
62 /* (From RFC1521 and draft-ietf-dnssec-secext-03.txt)
63    The following encoding technique is taken from RFC 1521 by Borenstein
64    and Freed.  It is reproduced here in a slightly edited form for
65    convenience.
66 
67    A 65-character subset of US-ASCII is used, enabling 6 bits to be
68    represented per printable character. (The extra 65th character, "=",
69    is used to signify a special processing function.)
70 
71    The encoding process represents 24-bit groups of input bits as output
72    strings of 4 encoded characters. Proceeding from left to right, a
73    24-bit input group is formed by concatenating 3 8-bit input groups.
74    These 24 bits are then treated as 4 concatenated 6-bit groups, each
75    of which is translated into a single digit in the base64 alphabet.
76 
77    Each 6-bit group is used as an index into an array of 64 printable
78    characters. The character referenced by the index is placed in the
79    output string.
80 
81                          Table 1: The Base64 Alphabet
82 
83       Value Encoding  Value Encoding  Value Encoding  Value Encoding
84           0 A            17 R            34 i            51 z
85           1 B            18 S            35 j            52 0
86           2 C            19 T            36 k            53 1
87           3 D            20 U            37 l            54 2
88           4 E            21 V            38 m            55 3
89           5 F            22 W            39 n            56 4
90           6 G            23 X            40 o            57 5
91           7 H            24 Y            41 p            58 6
92           8 I            25 Z            42 q            59 7
93           9 J            26 a            43 r            60 8
94          10 K            27 b            44 s            61 9
95          11 L            28 c            45 t            62 +
96          12 M            29 d            46 u            63 /
97          13 N            30 e            47 v
98          14 O            31 f            48 w         (pad) =
99          15 P            32 g            49 x
100          16 Q            33 h            50 y
101 
102    Special processing is performed if fewer than 24 bits are available
103    at the end of the data being encoded.  A full encoding quantum is
104    always completed at the end of a quantity.  When fewer than 24 input
105    bits are available in an input group, zero bits are added (on the
106    right) to form an integral number of 6-bit groups.  Padding at the
107    end of the data is performed using the '=' character.
108 
109    Since all base64 input is an integral number of octets, only the
110    following cases can arise:
111 
112        (1) the final quantum of encoding input is an integral
113            multiple of 24 bits; here, the final unit of encoded
114 	   output will be an integral multiple of 4 characters
115 	   with no "=" padding,
116        (2) the final quantum of encoding input is exactly 8 bits;
117            here, the final unit of encoded output will be two
118 	   characters followed by two "=" padding characters, or
119        (3) the final quantum of encoding input is exactly 16 bits;
120            here, the final unit of encoded output will be three
121 	   characters followed by one "=" padding character.
122    */
123 
124 /* skips all whitespace anywhere.
125    converts characters, four at a time, starting at (or after)
126    src from base - 64 numbers into three 8 bit bytes in the target area.
127    it returns the number of data bytes stored at the target, or -1 on error.
128  */
129 
130 static int b64rmap_initialized = 0;
131 static uint8_t b64rmap[256];
132 
133 static const uint8_t b64rmap_special = 0xf0;
134 static const uint8_t b64rmap_end = 0xfd;
135 static const uint8_t b64rmap_space = 0xfe;
136 static const uint8_t b64rmap_invalid = 0xff;
137 
138 /**
139  * Initializing the reverse map is not thread safe.
140  * Which is fine for NSD. For now...
141  **/
142 static void
b64_initialize_rmap()143 b64_initialize_rmap ()
144 {
145 	int i;
146 	char ch;
147 
148 	/* Null: end of string, stop parsing */
149 	b64rmap[0] = b64rmap_end;
150 
151 	for (i = 1; i < 256; ++i) {
152 		ch = (char)i;
153 		/* Whitespaces */
154 		if (isspace(ch))
155 			b64rmap[i] = b64rmap_space;
156 		/* Padding: stop parsing */
157 		else if (ch == Pad64)
158 			b64rmap[i] = b64rmap_end;
159 		/* Non-base64 char */
160 		else
161 			b64rmap[i] = b64rmap_invalid;
162 	}
163 
164 	/* Fill reverse mapping for base64 chars */
165 	for (i = 0; Base64[i] != '\0'; ++i)
166 		b64rmap[(uint8_t)Base64[i]] = i;
167 
168 	b64rmap_initialized = 1;
169 }
170 
171 static int
b64_pton_do(char const * src,uint8_t * target,size_t targsize)172 b64_pton_do(char const *src, uint8_t *target, size_t targsize)
173 {
174 	int tarindex, state, ch;
175 	uint8_t ofs;
176 
177 	state = 0;
178 	tarindex = 0;
179 
180 	while (1)
181 	{
182 		ch = *src++;
183 		ofs = b64rmap[ch];
184 
185 		if (ofs >= b64rmap_special) {
186 			/* Ignore whitespaces */
187 			if (ofs == b64rmap_space)
188 				continue;
189 			/* End of base64 characters */
190 			if (ofs == b64rmap_end)
191 				break;
192 			/* A non-base64 character. */
193 			return (-1);
194 		}
195 
196 		switch (state) {
197 		case 0:
198 			if ((size_t)tarindex >= targsize)
199 				return (-1);
200 			target[tarindex] = ofs << 2;
201 			state = 1;
202 			break;
203 		case 1:
204 			if ((size_t)tarindex + 1 >= targsize)
205 				return (-1);
206 			target[tarindex]   |=  ofs >> 4;
207 			target[tarindex+1]  = (ofs & 0x0f)
208 						<< 4 ;
209 			tarindex++;
210 			state = 2;
211 			break;
212 		case 2:
213 			if ((size_t)tarindex + 1 >= targsize)
214 				return (-1);
215 			target[tarindex]   |=  ofs >> 2;
216 			target[tarindex+1]  = (ofs & 0x03)
217 						<< 6;
218 			tarindex++;
219 			state = 3;
220 			break;
221 		case 3:
222 			if ((size_t)tarindex >= targsize)
223 				return (-1);
224 			target[tarindex] |= ofs;
225 			tarindex++;
226 			state = 0;
227 			break;
228 		default:
229 			abort();
230 		}
231 	}
232 
233 	/*
234 	 * We are done decoding Base-64 chars.  Let's see if we ended
235 	 * on a byte boundary, and/or with erroneous trailing characters.
236 	 */
237 
238 	if (ch == Pad64) {		/* We got a pad char. */
239 		ch = *src++;		/* Skip it, get next. */
240 		switch (state) {
241 		case 0:		/* Invalid = in first position */
242 		case 1:		/* Invalid = in second position */
243 			return (-1);
244 
245 		case 2:		/* Valid, means one byte of info */
246 			/* Skip any number of spaces. */
247 			for ((void)NULL; ch != '\0'; ch = *src++)
248 				if (b64rmap[ch] != b64rmap_space)
249 					break;
250 			/* Make sure there is another trailing = sign. */
251 			if (ch != Pad64)
252 				return (-1);
253 			ch = *src++;		/* Skip the = */
254 			/* Fall through to "single trailing =" case. */
255 			/* FALLTHROUGH */
256 
257 		case 3:		/* Valid, means two bytes of info */
258 			/*
259 			 * We know this char is an =.  Is there anything but
260 			 * whitespace after it?
261 			 */
262 			for ((void)NULL; ch != '\0'; ch = *src++)
263 				if (b64rmap[ch] != b64rmap_space)
264 					return (-1);
265 
266 			/*
267 			 * Now make sure for cases 2 and 3 that the "extra"
268 			 * bits that slopped past the last full byte were
269 			 * zeros.  If we don't check them, they become a
270 			 * subliminal channel.
271 			 */
272 			if (target[tarindex] != 0)
273 				return (-1);
274 		}
275 	} else {
276 		/*
277 		 * We ended by seeing the end of the string.  Make sure we
278 		 * have no partial bytes lying around.
279 		 */
280 		if (state != 0)
281 			return (-1);
282 	}
283 
284 	return (tarindex);
285 }
286 
287 
288 static int
b64_pton_len(char const * src)289 b64_pton_len(char const *src)
290 {
291 	int tarindex, state, ch;
292 	uint8_t ofs;
293 
294 	state = 0;
295 	tarindex = 0;
296 
297 	while (1)
298 	{
299 		ch = *src++;
300 		ofs = b64rmap[ch];
301 
302 		if (ofs >= b64rmap_special) {
303 			/* Ignore whitespaces */
304 			if (ofs == b64rmap_space)
305 				continue;
306 			/* End of base64 characters */
307 			if (ofs == b64rmap_end)
308 				break;
309 			/* A non-base64 character. */
310 			return (-1);
311 		}
312 
313 		switch (state) {
314 		case 0:
315 			state = 1;
316 			break;
317 		case 1:
318 			tarindex++;
319 			state = 2;
320 			break;
321 		case 2:
322 			tarindex++;
323 			state = 3;
324 			break;
325 		case 3:
326 			tarindex++;
327 			state = 0;
328 			break;
329 		default:
330 			abort();
331 		}
332 	}
333 
334 	/*
335 	 * We are done decoding Base-64 chars.  Let's see if we ended
336 	 * on a byte boundary, and/or with erroneous trailing characters.
337 	 */
338 
339 	if (ch == Pad64) {		/* We got a pad char. */
340 		ch = *src++;		/* Skip it, get next. */
341 		switch (state) {
342 		case 0:		/* Invalid = in first position */
343 		case 1:		/* Invalid = in second position */
344 			return (-1);
345 
346 		case 2:		/* Valid, means one byte of info */
347 			/* Skip any number of spaces. */
348 			for ((void)NULL; ch != '\0'; ch = *src++)
349 				if (b64rmap[ch] != b64rmap_space)
350 					break;
351 			/* Make sure there is another trailing = sign. */
352 			if (ch != Pad64)
353 				return (-1);
354 			ch = *src++;		/* Skip the = */
355 			/* Fall through to "single trailing =" case. */
356 			/* FALLTHROUGH */
357 
358 		case 3:		/* Valid, means two bytes of info */
359 			/*
360 			 * We know this char is an =.  Is there anything but
361 			 * whitespace after it?
362 			 */
363 			for ((void)NULL; ch != '\0'; ch = *src++)
364 				if (b64rmap[ch] != b64rmap_space)
365 					return (-1);
366 
367 		}
368 	} else {
369 		/*
370 		 * We ended by seeing the end of the string.  Make sure we
371 		 * have no partial bytes lying around.
372 		 */
373 		if (state != 0)
374 			return (-1);
375 	}
376 
377 	return (tarindex);
378 }
379 
380 
381 int
b64_pton(char const * src,uint8_t * target,size_t targsize)382 b64_pton(char const *src, uint8_t *target, size_t targsize)
383 {
384 	if (!b64rmap_initialized)
385 		b64_initialize_rmap ();
386 
387 	if (target)
388 		return b64_pton_do (src, target, targsize);
389 	else
390 		return b64_pton_len (src);
391 }
392