1/* base64.c -- routines to encode/decode base64 data */ 2/* $OpenLDAP$ */ 3/* This work is part of OpenLDAP Software <http://www.openldap.org/>. 4 * 5 * Copyright 1998-2011 The OpenLDAP Foundation. 6 * Portions Copyright 1998-2003 Kurt D. Zeilenga. 7 * Portions Copyright 1995 IBM Corporation. 8 * All rights reserved. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted only as authorized by the OpenLDAP 12 * Public License. 13 * 14 * A copy of this license is available in the file LICENSE in the 15 * top-level directory of the distribution or, alternatively, at 16 * <http://www.OpenLDAP.org/license.html>. 17 */ 18/* Portions Copyright (c) 1996, 1998 by Internet Software Consortium. 19 * 20 * Permission to use, copy, modify, and distribute this software for any 21 * purpose with or without fee is hereby granted, provided that the above 22 * copyright notice and this permission notice appear in all copies. 23 * 24 * THE SOFTWARE IS PROVIDED "AS IS" AND INTERNET SOFTWARE CONSORTIUM DISCLAIMS 25 * ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES 26 * OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL INTERNET SOFTWARE 27 * CONSORTIUM BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL 28 * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR 29 * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS 30 * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS 31 * SOFTWARE. 32 */ 33/* This work is based upon Base64 routines (developed by IBM) found 34 * Berkeley Internet Name Daemon (BIND) as distributed by ISC. They 35 * were adapted for inclusion in OpenLDAP Software by Kurt D. Zeilenga. 36 */ 37 38#include "portable.h" 39 40#include <ac/assert.h> 41#include <ac/stdlib.h> 42#include <ac/ctype.h> 43#include <ac/string.h> 44 45/* include socket.h to get sys/types.h and/or winsock2.h */ 46#include <ac/socket.h> 47 48#include "lutil.h" 49 50static const char Base64[] = 51 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; 52static const char Pad64 = '='; 53 54/* (From RFC1521 and draft-ietf-dnssec-secext-03.txt) 55 The following encoding technique is taken from RFC 1521 by Borenstein 56 and Freed. It is reproduced here in a slightly edited form for 57 convenience. 58 59 A 65-character subset of US-ASCII is used, enabling 6 bits to be 60 represented per printable character. (The extra 65th character, "=", 61 is used to signify a special processing function.) 62 63 The encoding process represents 24-bit groups of input bits as output 64 strings of 4 encoded characters. Proceeding from left to right, a 65 24-bit input group is formed by concatenating 3 8-bit input groups. 66 These 24 bits are then treated as 4 concatenated 6-bit groups, each 67 of which is translated into a single digit in the base64 alphabet. 68 69 Each 6-bit group is used as an index into an array of 64 printable 70 characters. The character referenced by the index is placed in the 71 output string. 72 73 Table 1: The Base64 Alphabet 74 75 Value Encoding Value Encoding Value Encoding Value Encoding 76 0 A 17 R 34 i 51 z 77 1 B 18 S 35 j 52 0 78 2 C 19 T 36 k 53 1 79 3 D 20 U 37 l 54 2 80 4 E 21 V 38 m 55 3 81 5 F 22 W 39 n 56 4 82 6 G 23 X 40 o 57 5 83 7 H 24 Y 41 p 58 6 84 8 I 25 Z 42 q 59 7 85 9 J 26 a 43 r 60 8 86 10 K 27 b 44 s 61 9 87 11 L 28 c 45 t 62 + 88 12 M 29 d 46 u 63 / 89 13 N 30 e 47 v 90 14 O 31 f 48 w (pad) = 91 15 P 32 g 49 x 92 16 Q 33 h 50 y 93 94 Special processing is performed if fewer than 24 bits are available 95 at the end of the data being encoded. A full encoding quantum is 96 always completed at the end of a quantity. When fewer than 24 input 97 bits are available in an input group, zero bits are added (on the 98 right) to form an integral number of 6-bit groups. Padding at the 99 end of the data is performed using the '=' character. 100 101 Since all base64 input is an integral number of octets, only the 102 ------------------------------------------------- 103 following cases can arise: 104 105 (1) the final quantum of encoding input is an integral 106 multiple of 24 bits; here, the final unit of encoded 107 output will be an integral multiple of 4 characters 108 with no "=" padding, 109 (2) the final quantum of encoding input is exactly 8 bits; 110 here, the final unit of encoded output will be two 111 characters followed by two "=" padding characters, or 112 (3) the final quantum of encoding input is exactly 16 bits; 113 here, the final unit of encoded output will be three 114 characters followed by one "=" padding character. 115 */ 116 117int 118lutil_b64_ntop( 119 u_char const *src, 120 size_t srclength, 121 char *target, 122 size_t targsize) 123{ 124 size_t datalength = 0; 125 u_char input[3]; 126 u_char output[4]; 127 size_t i; 128 129 while (2 < srclength) { 130 input[0] = *src++; 131 input[1] = *src++; 132 input[2] = *src++; 133 srclength -= 3; 134 135 output[0] = input[0] >> 2; 136 output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4); 137 output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6); 138 output[3] = input[2] & 0x3f; 139 assert(output[0] < 64); 140 assert(output[1] < 64); 141 assert(output[2] < 64); 142 assert(output[3] < 64); 143 144 if (datalength + 4 > targsize) 145 return (-1); 146 target[datalength++] = Base64[output[0]]; 147 target[datalength++] = Base64[output[1]]; 148 target[datalength++] = Base64[output[2]]; 149 target[datalength++] = Base64[output[3]]; 150 } 151 152 /* Now we worry about padding. */ 153 if (0 != srclength) { 154 /* Get what's left. */ 155 input[0] = input[1] = input[2] = '\0'; 156 for (i = 0; i < srclength; i++) 157 input[i] = *src++; 158 159 output[0] = input[0] >> 2; 160 output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4); 161 output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6); 162 assert(output[0] < 64); 163 assert(output[1] < 64); 164 assert(output[2] < 64); 165 166 if (datalength + 4 > targsize) 167 return (-1); 168 target[datalength++] = Base64[output[0]]; 169 target[datalength++] = Base64[output[1]]; 170 if (srclength == 1) 171 target[datalength++] = Pad64; 172 else 173 target[datalength++] = Base64[output[2]]; 174 target[datalength++] = Pad64; 175 } 176 if (datalength >= targsize) 177 return (-1); 178 target[datalength] = '\0'; /* Returned value doesn't count \0. */ 179 return (datalength); 180} 181 182/* skips all whitespace anywhere. 183 converts characters, four at a time, starting at (or after) 184 src from base - 64 numbers into three 8 bit bytes in the target area. 185 it returns the number of data bytes stored at the target, or -1 on error. 186 */ 187 188int 189lutil_b64_pton( 190 char const *src, 191 u_char *target, 192 size_t targsize) 193{ 194 int tarindex, state, ch; 195 char *pos; 196 197 state = 0; 198 tarindex = 0; 199 200 while ((ch = *src++) != '\0') { 201 if (isascii(ch) && isspace(ch)) /* Skip whitespace anywhere. */ 202 continue; 203 204 if (ch == Pad64) 205 break; 206 207 pos = strchr(Base64, ch); 208 if (pos == 0) /* A non-base64 character. */ 209 return (-1); 210 211 switch (state) { 212 case 0: 213 if (target) { 214 if ((size_t)tarindex >= targsize) 215 return (-1); 216 target[tarindex] = (pos - Base64) << 2; 217 } 218 state = 1; 219 break; 220 case 1: 221 if (target) { 222 if ((size_t)tarindex + 1 >= targsize) 223 return (-1); 224 target[tarindex] |= (pos - Base64) >> 4; 225 target[tarindex+1] = ((pos - Base64) & 0x0f) 226 << 4 ; 227 } 228 tarindex++; 229 state = 2; 230 break; 231 case 2: 232 if (target) { 233 if ((size_t)tarindex + 1 >= targsize) 234 return (-1); 235 target[tarindex] |= (pos - Base64) >> 2; 236 target[tarindex+1] = ((pos - Base64) & 0x03) 237 << 6; 238 } 239 tarindex++; 240 state = 3; 241 break; 242 case 3: 243 if (target) { 244 if ((size_t)tarindex >= targsize) 245 return (-1); 246 target[tarindex] |= (pos - Base64); 247 } 248 tarindex++; 249 state = 0; 250 break; 251 default: 252 abort(); 253 } 254 } 255 256 /* 257 * We are done decoding Base-64 chars. Let's see if we ended 258 * on a byte boundary, and/or with erroneous trailing characters. 259 */ 260 261 if (ch == Pad64) { /* We got a pad char. */ 262 ch = *src++; /* Skip it, get next. */ 263 switch (state) { 264 case 0: /* Invalid = in first position */ 265 case 1: /* Invalid = in second position */ 266 return (-1); 267 268 case 2: /* Valid, means one byte of info */ 269 /* Skip any number of spaces. */ 270 for ((void)NULL; ch != '\0'; ch = *src++) 271 if (! (isascii(ch) && isspace(ch))) 272 break; 273 /* Make sure there is another trailing = sign. */ 274 if (ch != Pad64) 275 return (-1); 276 ch = *src++; /* Skip the = */ 277 /* Fall through to "single trailing =" case. */ 278 /* FALLTHROUGH */ 279 280 case 3: /* Valid, means two bytes of info */ 281 /* 282 * We know this char is an =. Is there anything but 283 * whitespace after it? 284 */ 285 for ((void)NULL; ch != '\0'; ch = *src++) 286 if (! (isascii(ch) && isspace(ch))) 287 return (-1); 288 289 /* 290 * Now make sure for cases 2 and 3 that the "extra" 291 * bits that slopped past the last full byte were 292 * zeros. If we don't check them, they become a 293 * subliminal channel. 294 */ 295 if (target && target[tarindex] != 0) 296 return (-1); 297 } 298 } else { 299 /* 300 * We ended by seeing the end of the string. Make sure we 301 * have no partial bytes lying around. 302 */ 303 if (state != 0) 304 return (-1); 305 } 306 307 return (tarindex); 308} 309