1/* base64.c -- Encode binary data using printable characters. 2 Copyright (C) 1999, 2000, 2001, 2004, 2005, 2006, 2009, 2010 Free Software 3 Foundation, Inc. 4 5 This program is free software; you can redistribute it and/or modify 6 it under the terms of the GNU General Public License as published by 7 the Free Software Foundation; either version 3, or (at your option) 8 any later version. 9 10 This program is distributed in the hope that it will be useful, 11 but WITHOUT ANY WARRANTY; without even the implied warranty of 12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 13 GNU General Public License for more details. 14 15 You should have received a copy of the GNU General Public License 16 along with this program; if not, write to the Free Software Foundation, 17 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ 18 19/* Written by Simon Josefsson. Partially adapted from GNU MailUtils 20 * (mailbox/filter_trans.c, as of 2004-11-28). Improved by review 21 * from Paul Eggert, Bruno Haible, and Stepan Kasal. 22 * 23 * See also RFC 3548 <http://www.ietf.org/rfc/rfc3548.txt>. 24 * 25 * Be careful with error checking. Here is how you would typically 26 * use these functions: 27 * 28 * bool ok = base64_decode_alloc (in, inlen, &out, &outlen); 29 * if (!ok) 30 * FAIL: input was not valid base64 31 * if (out == NULL) 32 * FAIL: memory allocation error 33 * OK: data in OUT/OUTLEN 34 * 35 * size_t outlen = base64_encode_alloc (in, inlen, &out); 36 * if (out == NULL && outlen == 0 && inlen != 0) 37 * FAIL: input too long 38 * if (out == NULL) 39 * FAIL: memory allocation error 40 * OK: data in OUT/OUTLEN. 41 * 42 */ 43 44/* jude nagurney: GeoIP change, for providing proxy authorization */ 45/* #include <config.h> */ 46#define restrict __restrict 47 48/* Get prototype. */ 49#include "base64.h" 50 51/* Get malloc. */ 52#include <stdlib.h> 53 54/* Get UCHAR_MAX. */ 55#include <limits.h> 56 57#include <string.h> 58 59/* C89 compliant way to cast 'char' to 'unsigned char'. */ 60static inline unsigned char 61to_uchar (char ch) 62{ 63 return ch; 64} 65 66/* Base64 encode IN array of size INLEN into OUT array of size OUTLEN. 67 If OUTLEN is less than BASE64_LENGTH(INLEN), write as many bytes as 68 possible. If OUTLEN is larger than BASE64_LENGTH(INLEN), also zero 69 terminate the output buffer. */ 70void 71base64_encode (const char *restrict in, size_t inlen, 72 char *restrict out, size_t outlen) 73{ 74 static const char b64str[64] = 75 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; 76 77 while (inlen && outlen) 78 { 79 *out++ = b64str[(to_uchar (in[0]) >> 2) & 0x3f]; 80 if (!--outlen) 81 break; 82 *out++ = b64str[((to_uchar (in[0]) << 4) 83 + (--inlen ? to_uchar (in[1]) >> 4 : 0)) 84 & 0x3f]; 85 if (!--outlen) 86 break; 87 *out++ = 88 (inlen 89 ? b64str[((to_uchar (in[1]) << 2) 90 + (--inlen ? to_uchar (in[2]) >> 6 : 0)) 91 & 0x3f] 92 : '='); 93 if (!--outlen) 94 break; 95 *out++ = inlen ? b64str[to_uchar (in[2]) & 0x3f] : '='; 96 if (!--outlen) 97 break; 98 if (inlen) 99 inlen--; 100 if (inlen) 101 in += 3; 102 } 103 104 if (outlen) 105 *out = '\0'; 106} 107 108/* Allocate a buffer and store zero terminated base64 encoded data 109 from array IN of size INLEN, returning BASE64_LENGTH(INLEN), i.e., 110 the length of the encoded data, excluding the terminating zero. On 111 return, the OUT variable will hold a pointer to newly allocated 112 memory that must be deallocated by the caller. If output string 113 length would overflow, 0 is returned and OUT is set to NULL. If 114 memory allocation failed, OUT is set to NULL, and the return value 115 indicates length of the requested memory block, i.e., 116 BASE64_LENGTH(inlen) + 1. */ 117size_t 118base64_encode_alloc (const char *in, size_t inlen, char **out) 119{ 120 size_t outlen = 1 + BASE64_LENGTH (inlen); 121 122 /* Check for overflow in outlen computation. 123 * 124 * If there is no overflow, outlen >= inlen. 125 * 126 * If the operation (inlen + 2) overflows then it yields at most +1, so 127 * outlen is 0. 128 * 129 * If the multiplication overflows, we lose at least half of the 130 * correct value, so the result is < ((inlen + 2) / 3) * 2, which is 131 * less than (inlen + 2) * 0.66667, which is less than inlen as soon as 132 * (inlen > 4). 133 */ 134 if (inlen > outlen) 135 { 136 *out = NULL; 137 return 0; 138 } 139 140 *out = malloc (outlen); 141 if (!*out) 142 return outlen; 143 144 base64_encode (in, inlen, *out, outlen); 145 146 return outlen - 1; 147} 148 149/* With this approach this file works independent of the charset used 150 (think EBCDIC). However, it does assume that the characters in the 151 Base64 alphabet (A-Za-z0-9+/) are encoded in 0..255. POSIX 152 1003.1-2001 require that char and unsigned char are 8-bit 153 quantities, though, taking care of that problem. But this may be a 154 potential problem on non-POSIX C99 platforms. 155 156 IBM C V6 for AIX mishandles "#define B64(x) ...'x'...", so use "_" 157 as the formal parameter rather than "x". */ 158#define B64(_) \ 159 ((_) == 'A' ? 0 \ 160 : (_) == 'B' ? 1 \ 161 : (_) == 'C' ? 2 \ 162 : (_) == 'D' ? 3 \ 163 : (_) == 'E' ? 4 \ 164 : (_) == 'F' ? 5 \ 165 : (_) == 'G' ? 6 \ 166 : (_) == 'H' ? 7 \ 167 : (_) == 'I' ? 8 \ 168 : (_) == 'J' ? 9 \ 169 : (_) == 'K' ? 10 \ 170 : (_) == 'L' ? 11 \ 171 : (_) == 'M' ? 12 \ 172 : (_) == 'N' ? 13 \ 173 : (_) == 'O' ? 14 \ 174 : (_) == 'P' ? 15 \ 175 : (_) == 'Q' ? 16 \ 176 : (_) == 'R' ? 17 \ 177 : (_) == 'S' ? 18 \ 178 : (_) == 'T' ? 19 \ 179 : (_) == 'U' ? 20 \ 180 : (_) == 'V' ? 21 \ 181 : (_) == 'W' ? 22 \ 182 : (_) == 'X' ? 23 \ 183 : (_) == 'Y' ? 24 \ 184 : (_) == 'Z' ? 25 \ 185 : (_) == 'a' ? 26 \ 186 : (_) == 'b' ? 27 \ 187 : (_) == 'c' ? 28 \ 188 : (_) == 'd' ? 29 \ 189 : (_) == 'e' ? 30 \ 190 : (_) == 'f' ? 31 \ 191 : (_) == 'g' ? 32 \ 192 : (_) == 'h' ? 33 \ 193 : (_) == 'i' ? 34 \ 194 : (_) == 'j' ? 35 \ 195 : (_) == 'k' ? 36 \ 196 : (_) == 'l' ? 37 \ 197 : (_) == 'm' ? 38 \ 198 : (_) == 'n' ? 39 \ 199 : (_) == 'o' ? 40 \ 200 : (_) == 'p' ? 41 \ 201 : (_) == 'q' ? 42 \ 202 : (_) == 'r' ? 43 \ 203 : (_) == 's' ? 44 \ 204 : (_) == 't' ? 45 \ 205 : (_) == 'u' ? 46 \ 206 : (_) == 'v' ? 47 \ 207 : (_) == 'w' ? 48 \ 208 : (_) == 'x' ? 49 \ 209 : (_) == 'y' ? 50 \ 210 : (_) == 'z' ? 51 \ 211 : (_) == '0' ? 52 \ 212 : (_) == '1' ? 53 \ 213 : (_) == '2' ? 54 \ 214 : (_) == '3' ? 55 \ 215 : (_) == '4' ? 56 \ 216 : (_) == '5' ? 57 \ 217 : (_) == '6' ? 58 \ 218 : (_) == '7' ? 59 \ 219 : (_) == '8' ? 60 \ 220 : (_) == '9' ? 61 \ 221 : (_) == '+' ? 62 \ 222 : (_) == '/' ? 63 \ 223 : -1) 224 225static const signed char b64[0x100] = { 226 B64 (0), B64 (1), B64 (2), B64 (3), 227 B64 (4), B64 (5), B64 (6), B64 (7), 228 B64 (8), B64 (9), B64 (10), B64 (11), 229 B64 (12), B64 (13), B64 (14), B64 (15), 230 B64 (16), B64 (17), B64 (18), B64 (19), 231 B64 (20), B64 (21), B64 (22), B64 (23), 232 B64 (24), B64 (25), B64 (26), B64 (27), 233 B64 (28), B64 (29), B64 (30), B64 (31), 234 B64 (32), B64 (33), B64 (34), B64 (35), 235 B64 (36), B64 (37), B64 (38), B64 (39), 236 B64 (40), B64 (41), B64 (42), B64 (43), 237 B64 (44), B64 (45), B64 (46), B64 (47), 238 B64 (48), B64 (49), B64 (50), B64 (51), 239 B64 (52), B64 (53), B64 (54), B64 (55), 240 B64 (56), B64 (57), B64 (58), B64 (59), 241 B64 (60), B64 (61), B64 (62), B64 (63), 242 B64 (64), B64 (65), B64 (66), B64 (67), 243 B64 (68), B64 (69), B64 (70), B64 (71), 244 B64 (72), B64 (73), B64 (74), B64 (75), 245 B64 (76), B64 (77), B64 (78), B64 (79), 246 B64 (80), B64 (81), B64 (82), B64 (83), 247 B64 (84), B64 (85), B64 (86), B64 (87), 248 B64 (88), B64 (89), B64 (90), B64 (91), 249 B64 (92), B64 (93), B64 (94), B64 (95), 250 B64 (96), B64 (97), B64 (98), B64 (99), 251 B64 (100), B64 (101), B64 (102), B64 (103), 252 B64 (104), B64 (105), B64 (106), B64 (107), 253 B64 (108), B64 (109), B64 (110), B64 (111), 254 B64 (112), B64 (113), B64 (114), B64 (115), 255 B64 (116), B64 (117), B64 (118), B64 (119), 256 B64 (120), B64 (121), B64 (122), B64 (123), 257 B64 (124), B64 (125), B64 (126), B64 (127), 258 B64 (128), B64 (129), B64 (130), B64 (131), 259 B64 (132), B64 (133), B64 (134), B64 (135), 260 B64 (136), B64 (137), B64 (138), B64 (139), 261 B64 (140), B64 (141), B64 (142), B64 (143), 262 B64 (144), B64 (145), B64 (146), B64 (147), 263 B64 (148), B64 (149), B64 (150), B64 (151), 264 B64 (152), B64 (153), B64 (154), B64 (155), 265 B64 (156), B64 (157), B64 (158), B64 (159), 266 B64 (160), B64 (161), B64 (162), B64 (163), 267 B64 (164), B64 (165), B64 (166), B64 (167), 268 B64 (168), B64 (169), B64 (170), B64 (171), 269 B64 (172), B64 (173), B64 (174), B64 (175), 270 B64 (176), B64 (177), B64 (178), B64 (179), 271 B64 (180), B64 (181), B64 (182), B64 (183), 272 B64 (184), B64 (185), B64 (186), B64 (187), 273 B64 (188), B64 (189), B64 (190), B64 (191), 274 B64 (192), B64 (193), B64 (194), B64 (195), 275 B64 (196), B64 (197), B64 (198), B64 (199), 276 B64 (200), B64 (201), B64 (202), B64 (203), 277 B64 (204), B64 (205), B64 (206), B64 (207), 278 B64 (208), B64 (209), B64 (210), B64 (211), 279 B64 (212), B64 (213), B64 (214), B64 (215), 280 B64 (216), B64 (217), B64 (218), B64 (219), 281 B64 (220), B64 (221), B64 (222), B64 (223), 282 B64 (224), B64 (225), B64 (226), B64 (227), 283 B64 (228), B64 (229), B64 (230), B64 (231), 284 B64 (232), B64 (233), B64 (234), B64 (235), 285 B64 (236), B64 (237), B64 (238), B64 (239), 286 B64 (240), B64 (241), B64 (242), B64 (243), 287 B64 (244), B64 (245), B64 (246), B64 (247), 288 B64 (248), B64 (249), B64 (250), B64 (251), 289 B64 (252), B64 (253), B64 (254), B64 (255) 290}; 291 292#if UCHAR_MAX == 255 293# define uchar_in_range(c) true 294#else 295# define uchar_in_range(c) ((c) <= 255) 296#endif 297 298/* Return true if CH is a character from the Base64 alphabet, and 299 false otherwise. Note that '=' is padding and not considered to be 300 part of the alphabet. */ 301bool 302isbase64 (char ch) 303{ 304 return uchar_in_range (to_uchar (ch)) && 0 <= b64[to_uchar (ch)]; 305} 306 307/* Initialize decode-context buffer, CTX. */ 308void 309base64_decode_ctx_init (struct base64_decode_context *ctx) 310{ 311 ctx->i = 0; 312} 313 314/* If CTX->i is 0 or 4, there are four or more bytes in [*IN..IN_END), and 315 none of those four is a newline, then return *IN. Otherwise, copy up to 316 4 - CTX->i non-newline bytes from that range into CTX->buf, starting at 317 index CTX->i and setting CTX->i to reflect the number of bytes copied, 318 and return CTX->buf. In either case, advance *IN to point to the byte 319 after the last one processed, and set *N_NON_NEWLINE to the number of 320 verified non-newline bytes accessible through the returned pointer. */ 321static inline char * 322get_4 (struct base64_decode_context *ctx, 323 char const *restrict *in, char const *restrict in_end, 324 size_t *n_non_newline) 325{ 326 if (ctx->i == 4) 327 ctx->i = 0; 328 329 if (ctx->i == 0) 330 { 331 char const *t = *in; 332 if (4 <= in_end - *in && memchr (t, '\n', 4) == NULL) 333 { 334 /* This is the common case: no newline. */ 335 *in += 4; 336 *n_non_newline = 4; 337 return (char *) t; 338 } 339 } 340 341 { 342 /* Copy non-newline bytes into BUF. */ 343 char const *p = *in; 344 while (p < in_end) 345 { 346 char c = *p++; 347 if (c != '\n') 348 { 349 ctx->buf[ctx->i++] = c; 350 if (ctx->i == 4) 351 break; 352 } 353 } 354 355 *in = p; 356 *n_non_newline = ctx->i; 357 return ctx->buf; 358 } 359} 360 361#define return_false \ 362 do \ 363 { \ 364 *outp = out; \ 365 return false; \ 366 } \ 367 while (false) 368 369/* Decode up to four bytes of base64-encoded data, IN, of length INLEN 370 into the output buffer, *OUT, of size *OUTLEN bytes. Return true if 371 decoding is successful, false otherwise. If *OUTLEN is too small, 372 as many bytes as possible are written to *OUT. On return, advance 373 *OUT to point to the byte after the last one written, and decrement 374 *OUTLEN to reflect the number of bytes remaining in *OUT. */ 375static inline bool 376decode_4 (char const *restrict in, size_t inlen, 377 char *restrict *outp, size_t *outleft) 378{ 379 char *out = *outp; 380 if (inlen < 2) 381 return false; 382 383 if (!isbase64 (in[0]) || !isbase64 (in[1])) 384 return false; 385 386 if (*outleft) 387 { 388 *out++ = ((b64[to_uchar (in[0])] << 2) 389 | (b64[to_uchar (in[1])] >> 4)); 390 --*outleft; 391 } 392 393 if (inlen == 2) 394 return_false; 395 396 if (in[2] == '=') 397 { 398 if (inlen != 4) 399 return_false; 400 401 if (in[3] != '=') 402 return_false; 403 } 404 else 405 { 406 if (!isbase64 (in[2])) 407 return_false; 408 409 if (*outleft) 410 { 411 *out++ = (((b64[to_uchar (in[1])] << 4) & 0xf0) 412 | (b64[to_uchar (in[2])] >> 2)); 413 --*outleft; 414 } 415 416 if (inlen == 3) 417 return_false; 418 419 if (in[3] == '=') 420 { 421 if (inlen != 4) 422 return_false; 423 } 424 else 425 { 426 if (!isbase64 (in[3])) 427 return_false; 428 429 if (*outleft) 430 { 431 *out++ = (((b64[to_uchar (in[2])] << 6) & 0xc0) 432 | b64[to_uchar (in[3])]); 433 --*outleft; 434 } 435 } 436 } 437 438 *outp = out; 439 return true; 440} 441 442/* Decode base64-encoded input array IN of length INLEN to output array 443 OUT that can hold *OUTLEN bytes. The input data may be interspersed 444 with newlines. Return true if decoding was successful, i.e. if the 445 input was valid base64 data, false otherwise. If *OUTLEN is too 446 small, as many bytes as possible will be written to OUT. On return, 447 *OUTLEN holds the length of decoded bytes in OUT. Note that as soon 448 as any non-alphabet, non-newline character is encountered, decoding 449 is stopped and false is returned. If INLEN is zero, then process 450 only whatever data is stored in CTX. 451 452 Initially, CTX must have been initialized via base64_decode_ctx_init. 453 Subsequent calls to this function must reuse whatever state is recorded 454 in that buffer. It is necessary for when a quadruple of base64 input 455 bytes spans two input buffers. 456 457 If CTX is NULL then newlines are treated as garbage and the input 458 buffer is processed as a unit. */ 459 460bool 461base64_decode_ctx (struct base64_decode_context *ctx, 462 const char *restrict in, size_t inlen, 463 char *restrict out, size_t *outlen) 464{ 465 size_t outleft = *outlen; 466 bool ignore_newlines = ctx != NULL; 467 bool flush_ctx = false; 468 unsigned int ctx_i = 0; 469 470 if (ignore_newlines) 471 { 472 ctx_i = ctx->i; 473 flush_ctx = inlen == 0; 474 } 475 476 477 while (true) 478 { 479 size_t outleft_save = outleft; 480 if (ctx_i == 0 && !flush_ctx) 481 { 482 while (true) 483 { 484 /* Save a copy of outleft, in case we need to re-parse this 485 block of four bytes. */ 486 outleft_save = outleft; 487 if (!decode_4 (in, inlen, &out, &outleft)) 488 break; 489 490 in += 4; 491 inlen -= 4; 492 } 493 } 494 495 if (inlen == 0 && !flush_ctx) 496 break; 497 498 /* Handle the common case of 72-byte wrapped lines. 499 This also handles any other multiple-of-4-byte wrapping. */ 500 if (inlen && *in == '\n' && ignore_newlines) 501 { 502 ++in; 503 --inlen; 504 continue; 505 } 506 507 /* Restore OUT and OUTLEFT. */ 508 out -= outleft_save - outleft; 509 outleft = outleft_save; 510 511 { 512 char const *in_end = in + inlen; 513 char const *non_nl; 514 515 if (ignore_newlines) 516 non_nl = get_4 (ctx, &in, in_end, &inlen); 517 else 518 non_nl = in; /* Might have nl in this case. */ 519 520 /* If the input is empty or consists solely of newlines (0 non-newlines), 521 then we're done. Likewise if there are fewer than 4 bytes when not 522 flushing context and not treating newlines as garbage. */ 523 if (inlen == 0 || (inlen < 4 && !flush_ctx && ignore_newlines)) 524 { 525 inlen = 0; 526 break; 527 } 528 if (!decode_4 (non_nl, inlen, &out, &outleft)) 529 break; 530 531 inlen = in_end - in; 532 } 533 } 534 535 *outlen -= outleft; 536 537 return inlen == 0; 538} 539 540/* Allocate an output buffer in *OUT, and decode the base64 encoded 541 data stored in IN of size INLEN to the *OUT buffer. On return, the 542 size of the decoded data is stored in *OUTLEN. OUTLEN may be NULL, 543 if the caller is not interested in the decoded length. *OUT may be 544 NULL to indicate an out of memory error, in which case *OUTLEN 545 contains the size of the memory block needed. The function returns 546 true on successful decoding and memory allocation errors. (Use the 547 *OUT and *OUTLEN parameters to differentiate between successful 548 decoding and memory error.) The function returns false if the 549 input was invalid, in which case *OUT is NULL and *OUTLEN is 550 undefined. */ 551bool 552base64_decode_alloc_ctx (struct base64_decode_context *ctx, 553 const char *in, size_t inlen, char **out, 554 size_t *outlen) 555{ 556 /* This may allocate a few bytes too many, depending on input, 557 but it's not worth the extra CPU time to compute the exact size. 558 The exact size is 3 * inlen / 4, minus 1 if the input ends 559 with "=" and minus another 1 if the input ends with "==". 560 Dividing before multiplying avoids the possibility of overflow. */ 561 size_t needlen = 3 * (inlen / 4) + 2; 562 563 *out = malloc (needlen); 564 if (!*out) 565 return true; 566 567 if (!base64_decode_ctx (ctx, in, inlen, *out, &needlen)) 568 { 569 free (*out); 570 *out = NULL; 571 return false; 572 } 573 574 if (outlen) 575 *outlen = needlen; 576 577 return true; 578} 579