gl/base64.c

   1 /* base64.c -- Encode binary data using printable characters.
   2    Copyright (C) 1999-2001, 2004-2006, 2009-2012 Free Software Foundation, Inc.
   3
   4    This program is free software; you can redistribute it and/or modify
   5    it under the terms of the GNU General Public License as published by
   6    the Free Software Foundation; either version 3, or (at your option)
   7    any later version.
   8
   9    This program is distributed in the hope that it will be useful,
  10    but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12    GNU General Public License for more details.
  13
  14    You should have received a copy of the GNU General Public License
  15    along with this program; if not, see <http://www.gnu.org/licenses/>.  */
  16
  17 /* Written by Simon Josefsson.  Partially adapted from GNU MailUtils
  18  * (mailbox/filter_trans.c, as of 2004-11-28).  Improved by review
  19  * from Paul Eggert, Bruno Haible, and Stepan Kasal.
  20  *
  21  * See also RFC 4648 <http://www.ietf.org/rfc/rfc4648.txt>.
  22  *
  23  * Be careful with error checking.  Here is how you would typically
  24  * use these functions:
  25  *
  26  * bool ok = base64_decode_alloc (in, inlen, &out, &outlen);
  27  * if (!ok)
  28  *   FAIL: input was not valid base64
  29  * if (out == NULL)
  30  *   FAIL: memory allocation error
  31  * OK: data in OUT/OUTLEN
  32  *
  33  * size_t outlen = base64_encode_alloc (in, inlen, &out);
  34  * if (out == NULL && outlen == 0 && inlen != 0)
  35  *   FAIL: input too long
  36  * if (out == NULL)
  37  *   FAIL: memory allocation error
  38  * OK: data in OUT/OUTLEN.
  39  *
  40  */
  41
  42 #include <config.h>
  43
  44 /* Get prototype. */
  45 #include "base64.h"
  46
  47 /* Get malloc. */
  48 #include <stdlib.h>
  49
  50 /* Get UCHAR_MAX. */
  51 #include <limits.h>
  52
  53 #include <string.h>
  54
  55 /* C89 compliant way to cast 'char' to 'unsigned char'. */
  56 static inline unsigned char
  57 to_uchar (char ch)
  58 {
  59   return ch;
  60 }
  61
  62 /* Base64 encode IN array of size INLEN into OUT array of size OUTLEN.
  63    If OUTLEN is less than BASE64_LENGTH(INLEN), write as many bytes as
  64    possible.  If OUTLEN is larger than BASE64_LENGTH(INLEN), also zero
  65    terminate the output buffer. */
  66 void
  67 base64_encode (const char *restrict in, size_t inlen,
  68                char *restrict out, size_t outlen)
  69 {
  70   static const char b64str[64] =
  71     "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
  72
  73   while (inlen && outlen)
  74     {
  75       *out++ = b64str[(to_uchar (in[0]) >> 2) & 0x3f];
  76       if (!--outlen)
  77         break;
  78       *out++ = b64str[((to_uchar (in[0]) << 4)
  79                        + (--inlen ? to_uchar (in[1]) >> 4 : 0))
  80                       & 0x3f];
  81       if (!--outlen)
  82         break;
  83       *out++ =
  84         (inlen
  85          ? b64str[((to_uchar (in[1]) << 2)
  86                    + (--inlen ? to_uchar (in[2]) >> 6 : 0))
  87                   & 0x3f]
  88          : '=');
  89       if (!--outlen)
  90         break;
  91       *out++ = inlen ? b64str[to_uchar (in[2]) & 0x3f] : '=';
  92       if (!--outlen)
  93         break;
  94       if (inlen)
  95         inlen--;
  96       if (inlen)
  97         in += 3;
  98     }
  99
 100   if (outlen)
 101     *out = '\0';
 102 }
 103
 104 /* Allocate a buffer and store zero terminated base64 encoded data
 105    from array IN of size INLEN, returning BASE64_LENGTH(INLEN), i.e.,
 106    the length of the encoded data, excluding the terminating zero.  On
 107    return, the OUT variable will hold a pointer to newly allocated
 108    memory that must be deallocated by the caller.  If output string
 109    length would overflow, 0 is returned and OUT is set to NULL.  If
 110    memory allocation failed, OUT is set to NULL, and the return value
 111    indicates length of the requested memory block, i.e.,
 112    BASE64_LENGTH(inlen) + 1. */
 113 size_t
 114 base64_encode_alloc (const char *in, size_t inlen, char **out)
 115 {
 116   size_t outlen = 1 + BASE64_LENGTH (inlen);
 117
 118   /* Check for overflow in outlen computation.
 119    *
 120    * If there is no overflow, outlen >= inlen.
 121    *
 122    * If the operation (inlen + 2) overflows then it yields at most +1, so
 123    * outlen is 0.
 124    *
 125    * If the multiplication overflows, we lose at least half of the
 126    * correct value, so the result is < ((inlen + 2) / 3) * 2, which is
 127    * less than (inlen + 2) * 0.66667, which is less than inlen as soon as
 128    * (inlen > 4).
 129    */
 130   if (inlen > outlen)
 131     {
 132       *out = NULL;
 133       return 0;
 134     }
 135
 136   *out = malloc (outlen);
 137   if (!*out)
 138     return outlen;
 139
 140   base64_encode (in, inlen, *out, outlen);
 141
 142   return outlen - 1;
 143 }
 144
 145 /* With this approach this file works independent of the charset used
 146    (think EBCDIC).  However, it does assume that the characters in the
 147    Base64 alphabet (A-Za-z0-9+/) are encoded in 0..255.  POSIX
 148    1003.1-2001 require that char and unsigned char are 8-bit
 149    quantities, though, taking care of that problem.  But this may be a
 150    potential problem on non-POSIX C99 platforms.
 151
 152    IBM C V6 for AIX mishandles "#define B64(x) ...'x'...", so use "_"
 153    as the formal parameter rather than "x".  */
 154 #define B64(_)                                  \
 155   ((_) == 'A' ? 0                               \
 156    : (_) == 'B' ? 1                             \
 157    : (_) == 'C' ? 2                             \
 158    : (_) == 'D' ? 3                             \
 159    : (_) == 'E' ? 4                             \
 160    : (_) == 'F' ? 5                             \
 161    : (_) == 'G' ? 6                             \
 162    : (_) == 'H' ? 7                             \
 163    : (_) == 'I' ? 8                             \
 164    : (_) == 'J' ? 9                             \
 165    : (_) == 'K' ? 10                            \
 166    : (_) == 'L' ? 11                            \
 167    : (_) == 'M' ? 12                            \
 168    : (_) == 'N' ? 13                            \
 169    : (_) == 'O' ? 14                            \
 170    : (_) == 'P' ? 15                            \
 171    : (_) == 'Q' ? 16                            \
 172    : (_) == 'R' ? 17                            \
 173    : (_) == 'S' ? 18                            \
 174    : (_) == 'T' ? 19                            \
 175    : (_) == 'U' ? 20                            \
 176    : (_) == 'V' ? 21                            \
 177    : (_) == 'W' ? 22                            \
 178    : (_) == 'X' ? 23                            \
 179    : (_) == 'Y' ? 24                            \
 180    : (_) == 'Z' ? 25                            \
 181    : (_) == 'a' ? 26                            \
 182    : (_) == 'b' ? 27                            \
 183    : (_) == 'c' ? 28                            \
 184    : (_) == 'd' ? 29                            \
 185    : (_) == 'e' ? 30                            \
 186    : (_) == 'f' ? 31                            \
 187    : (_) == 'g' ? 32                            \
 188    : (_) == 'h' ? 33                            \
 189    : (_) == 'i' ? 34                            \
 190    : (_) == 'j' ? 35                            \
 191    : (_) == 'k' ? 36                            \
 192    : (_) == 'l' ? 37                            \
 193    : (_) == 'm' ? 38                            \
 194    : (_) == 'n' ? 39                            \
 195    : (_) == 'o' ? 40                            \
 196    : (_) == 'p' ? 41                            \
 197    : (_) == 'q' ? 42                            \
 198    : (_) == 'r' ? 43                            \
 199    : (_) == 's' ? 44                            \
 200    : (_) == 't' ? 45                            \
 201    : (_) == 'u' ? 46                            \
 202    : (_) == 'v' ? 47                            \
 203    : (_) == 'w' ? 48                            \
 204    : (_) == 'x' ? 49                            \
 205    : (_) == 'y' ? 50                            \
 206    : (_) == 'z' ? 51                            \
 207    : (_) == '0' ? 52                            \
 208    : (_) == '1' ? 53                            \
 209    : (_) == '2' ? 54                            \
 210    : (_) == '3' ? 55                            \
 211    : (_) == '4' ? 56                            \
 212    : (_) == '5' ? 57                            \
 213    : (_) == '6' ? 58                            \
 214    : (_) == '7' ? 59                            \
 215    : (_) == '8' ? 60                            \
 216    : (_) == '9' ? 61                            \
 217    : (_) == '+' ? 62                            \
 218    : (_) == '/' ? 63                            \
 219    : -1)
 220
 221 static const signed char b64[0x100] = {
 222   B64 (0), B64 (1), B64 (2), B64 (3),
 223   B64 (4), B64 (5), B64 (6), B64 (7),
 224   B64 (8), B64 (9), B64 (10), B64 (11),
 225   B64 (12), B64 (13), B64 (14), B64 (15),
 226   B64 (16), B64 (17), B64 (18), B64 (19),
 227   B64 (20), B64 (21), B64 (22), B64 (23),
 228   B64 (24), B64 (25), B64 (26), B64 (27),
 229   B64 (28), B64 (29), B64 (30), B64 (31),
 230   B64 (32), B64 (33), B64 (34), B64 (35),
 231   B64 (36), B64 (37), B64 (38), B64 (39),
 232   B64 (40), B64 (41), B64 (42), B64 (43),
 233   B64 (44), B64 (45), B64 (46), B64 (47),
 234   B64 (48), B64 (49), B64 (50), B64 (51),
 235   B64 (52), B64 (53), B64 (54), B64 (55),
 236   B64 (56), B64 (57), B64 (58), B64 (59),
 237   B64 (60), B64 (61), B64 (62), B64 (63),
 238   B64 (64), B64 (65), B64 (66), B64 (67),
 239   B64 (68), B64 (69), B64 (70), B64 (71),
 240   B64 (72), B64 (73), B64 (74), B64 (75),
 241   B64 (76), B64 (77), B64 (78), B64 (79),
 242   B64 (80), B64 (81), B64 (82), B64 (83),
 243   B64 (84), B64 (85), B64 (86), B64 (87),
 244   B64 (88), B64 (89), B64 (90), B64 (91),
 245   B64 (92), B64 (93), B64 (94), B64 (95),
 246   B64 (96), B64 (97), B64 (98), B64 (99),
 247   B64 (100), B64 (101), B64 (102), B64 (103),
 248   B64 (104), B64 (105), B64 (106), B64 (107),
 249   B64 (108), B64 (109), B64 (110), B64 (111),
 250   B64 (112), B64 (113), B64 (114), B64 (115),
 251   B64 (116), B64 (117), B64 (118), B64 (119),
 252   B64 (120), B64 (121), B64 (122), B64 (123),
 253   B64 (124), B64 (125), B64 (126), B64 (127),
 254   B64 (128), B64 (129), B64 (130), B64 (131),
 255   B64 (132), B64 (133), B64 (134), B64 (135),
 256   B64 (136), B64 (137), B64 (138), B64 (139),
 257   B64 (140), B64 (141), B64 (142), B64 (143),
 258   B64 (144), B64 (145), B64 (146), B64 (147),
 259   B64 (148), B64 (149), B64 (150), B64 (151),
 260   B64 (152), B64 (153), B64 (154), B64 (155),
 261   B64 (156), B64 (157), B64 (158), B64 (159),
 262   B64 (160), B64 (161), B64 (162), B64 (163),
 263   B64 (164), B64 (165), B64 (166), B64 (167),
 264   B64 (168), B64 (169), B64 (170), B64 (171),
 265   B64 (172), B64 (173), B64 (174), B64 (175),
 266   B64 (176), B64 (177), B64 (178), B64 (179),
 267   B64 (180), B64 (181), B64 (182), B64 (183),
 268   B64 (184), B64 (185), B64 (186), B64 (187),
 269   B64 (188), B64 (189), B64 (190), B64 (191),
 270   B64 (192), B64 (193), B64 (194), B64 (195),
 271   B64 (196), B64 (197), B64 (198), B64 (199),
 272   B64 (200), B64 (201), B64 (202), B64 (203),
 273   B64 (204), B64 (205), B64 (206), B64 (207),
 274   B64 (208), B64 (209), B64 (210), B64 (211),
 275   B64 (212), B64 (213), B64 (214), B64 (215),
 276   B64 (216), B64 (217), B64 (218), B64 (219),
 277   B64 (220), B64 (221), B64 (222), B64 (223),
 278   B64 (224), B64 (225), B64 (226), B64 (227),
 279   B64 (228), B64 (229), B64 (230), B64 (231),
 280   B64 (232), B64 (233), B64 (234), B64 (235),
 281   B64 (236), B64 (237), B64 (238), B64 (239),
 282   B64 (240), B64 (241), B64 (242), B64 (243),
 283   B64 (244), B64 (245), B64 (246), B64 (247),
 284   B64 (248), B64 (249), B64 (250), B64 (251),
 285   B64 (252), B64 (253), B64 (254), B64 (255)
 286 };
 287
 288 #if UCHAR_MAX == 255
 289 # define uchar_in_range(c) true
 290 #else
 291 # define uchar_in_range(c) ((c) <= 255)
 292 #endif
 293
 294 /* Return true if CH is a character from the Base64 alphabet, and
 295    false otherwise.  Note that '=' is padding and not considered to be
 296    part of the alphabet.  */
 297 bool
 298 isbase64 (char ch)
 299 {
 300   return uchar_in_range (to_uchar (ch)) && 0 <= b64[to_uchar (ch)];
 301 }
 302
 303 /* Initialize decode-context buffer, CTX.  */
 304 void
 305 base64_decode_ctx_init (struct base64_decode_context *ctx)
 306 {
 307   ctx->i = 0;
 308 }
 309
 310 /* If CTX->i is 0 or 4, there are four or more bytes in [*IN..IN_END), and
 311    none of those four is a newline, then return *IN.  Otherwise, copy up to
 312    4 - CTX->i non-newline bytes from that range into CTX->buf, starting at
 313    index CTX->i and setting CTX->i to reflect the number of bytes copied,
 314    and return CTX->buf.  In either case, advance *IN to point to the byte
 315    after the last one processed, and set *N_NON_NEWLINE to the number of
 316    verified non-newline bytes accessible through the returned pointer.  */
 317 static inline char *
 318 get_4 (struct base64_decode_context *ctx,
 319        char const *restrict *in, char const *restrict in_end,
 320        size_t *n_non_newline)
 321 {
 322   if (ctx->i == 4)
 323     ctx->i = 0;
 324
 325   if (ctx->i == 0)
 326     {
 327       char const *t = *in;
 328       if (4 <= in_end - *in && memchr (t, '\n', 4) == NULL)
 329         {
 330           /* This is the common case: no newline.  */
 331           *in += 4;
 332           *n_non_newline = 4;
 333           return (char *) t;
 334         }
 335     }
 336
 337   {
 338     /* Copy non-newline bytes into BUF.  */
 339     char const *p = *in;
 340     while (p < in_end)
 341       {
 342         char c = *p++;
 343         if (c != '\n')
 344           {
 345             ctx->buf[ctx->i++] = c;
 346             if (ctx->i == 4)
 347               break;
 348           }
 349       }
 350
 351     *in = p;
 352     *n_non_newline = ctx->i;
 353     return ctx->buf;
 354   }
 355 }
 356
 357 #define return_false                            \
 358   do                                            \
 359     {                                           \
 360       *outp = out;                              \
 361       return false;                             \
 362     }                                           \
 363   while (false)
 364
 365 /* Decode up to four bytes of base64-encoded data, IN, of length INLEN
 366    into the output buffer, *OUT, of size *OUTLEN bytes.  Return true if
 367    decoding is successful, false otherwise.  If *OUTLEN is too small,
 368    as many bytes as possible are written to *OUT.  On return, advance
 369    *OUT to point to the byte after the last one written, and decrement
 370    *OUTLEN to reflect the number of bytes remaining in *OUT.  */
 371 static inline bool
 372 decode_4 (char const *restrict in, size_t inlen,
 373           char *restrict *outp, size_t *outleft)
 374 {
 375   char *out = *outp;
 376   if (inlen < 2)
 377     return false;
 378
 379   if (!isbase64 (in[0]) || !isbase64 (in[1]))
 380     return false;
 381
 382   if (*outleft)
 383     {
 384       *out++ = ((b64[to_uchar (in[0])] << 2)
 385                 | (b64[to_uchar (in[1])] >> 4));
 386       --*outleft;
 387     }
 388
 389   if (inlen == 2)
 390     return_false;
 391
 392   if (in[2] == '=')
 393     {
 394       if (inlen != 4)
 395         return_false;
 396
 397       if (in[3] != '=')
 398         return_false;
 399     }
 400   else
 401     {
 402       if (!isbase64 (in[2]))
 403         return_false;
 404
 405       if (*outleft)
 406         {
 407           *out++ = (((b64[to_uchar (in[1])] << 4) & 0xf0)
 408                     | (b64[to_uchar (in[2])] >> 2));
 409           --*outleft;
 410         }
 411
 412       if (inlen == 3)
 413         return_false;
 414
 415       if (in[3] == '=')
 416         {
 417           if (inlen != 4)
 418             return_false;
 419         }
 420       else
 421         {
 422           if (!isbase64 (in[3]))
 423             return_false;
 424
 425           if (*outleft)
 426             {
 427               *out++ = (((b64[to_uchar (in[2])] << 6) & 0xc0)
 428                         | b64[to_uchar (in[3])]);
 429               --*outleft;
 430             }
 431         }
 432     }
 433
 434   *outp = out;
 435   return true;
 436 }
 437
 438 /* Decode base64-encoded input array IN of length INLEN to output array
 439    OUT that can hold *OUTLEN bytes.  The input data may be interspersed
 440    with newlines.  Return true if decoding was successful, i.e. if the
 441    input was valid base64 data, false otherwise.  If *OUTLEN is too
 442    small, as many bytes as possible will be written to OUT.  On return,
 443    *OUTLEN holds the length of decoded bytes in OUT.  Note that as soon
 444    as any non-alphabet, non-newline character is encountered, decoding
 445    is stopped and false is returned.  If INLEN is zero, then process
 446    only whatever data is stored in CTX.
 447
 448    Initially, CTX must have been initialized via base64_decode_ctx_init.
 449    Subsequent calls to this function must reuse whatever state is recorded
 450    in that buffer.  It is necessary for when a quadruple of base64 input
 451    bytes spans two input buffers.
 452
 453    If CTX is NULL then newlines are treated as garbage and the input
 454    buffer is processed as a unit.  */
 455
 456 bool
 457 base64_decode_ctx (struct base64_decode_context *ctx,
 458                    const char *restrict in, size_t inlen,
 459                    char *restrict out, size_t *outlen)
 460 {
 461   size_t outleft = *outlen;
 462   bool ignore_newlines = ctx != NULL;
 463   bool flush_ctx = false;
 464   unsigned int ctx_i = 0;
 465
 466   if (ignore_newlines)
 467     {
 468       ctx_i = ctx->i;
 469       flush_ctx = inlen == 0;
 470     }
 471
 472
 473   while (true)
 474     {
 475       size_t outleft_save = outleft;
 476       if (ctx_i == 0 && !flush_ctx)
 477         {
 478           while (true)
 479             {
 480               /* Save a copy of outleft, in case we need to re-parse this
 481                  block of four bytes.  */
 482               outleft_save = outleft;
 483               if (!decode_4 (in, inlen, &out, &outleft))
 484                 break;
 485
 486               in += 4;
 487               inlen -= 4;
 488             }
 489         }
 490
 491       if (inlen == 0 && !flush_ctx)
 492         break;
 493
 494       /* Handle the common case of 72-byte wrapped lines.
 495          This also handles any other multiple-of-4-byte wrapping.  */
 496       if (inlen && *in == '\n' && ignore_newlines)
 497         {
 498           ++in;
 499           --inlen;
 500           continue;
 501         }
 502
 503       /* Restore OUT and OUTLEFT.  */
 504       out -= outleft_save - outleft;
 505       outleft = outleft_save;
 506
 507       {
 508         char const *in_end = in + inlen;
 509         char const *non_nl;
 510
 511         if (ignore_newlines)
 512           non_nl = get_4 (ctx, &in, in_end, &inlen);
 513         else
 514           non_nl = in;  /* Might have nl in this case. */
 515
 516         /* If the input is empty or consists solely of newlines (0 non-newlines),
 517            then we're done.  Likewise if there are fewer than 4 bytes when not
 518            flushing context and not treating newlines as garbage.  */
 519         if (inlen == 0 || (inlen < 4 && !flush_ctx && ignore_newlines))
 520           {
 521             inlen = 0;
 522             break;
 523           }
 524         if (!decode_4 (non_nl, inlen, &out, &outleft))
 525           break;
 526
 527         inlen = in_end - in;
 528       }
 529     }
 530
 531   *outlen -= outleft;
 532
 533   return inlen == 0;
 534 }
 535
 536 /* Allocate an output buffer in *OUT, and decode the base64 encoded
 537    data stored in IN of size INLEN to the *OUT buffer.  On return, the
 538    size of the decoded data is stored in *OUTLEN.  OUTLEN may be NULL,
 539    if the caller is not interested in the decoded length.  *OUT may be
 540    NULL to indicate an out of memory error, in which case *OUTLEN
 541    contains the size of the memory block needed.  The function returns
 542    true on successful decoding and memory allocation errors.  (Use the
 543    *OUT and *OUTLEN parameters to differentiate between successful
 544    decoding and memory error.)  The function returns false if the
 545    input was invalid, in which case *OUT is NULL and *OUTLEN is
 546    undefined. */
 547 bool
 548 base64_decode_alloc_ctx (struct base64_decode_context *ctx,
 549                          const char *in, size_t inlen, char **out,
 550                          size_t *outlen)
 551 {
 552   /* This may allocate a few bytes too many, depending on input,
 553      but it's not worth the extra CPU time to compute the exact size.
 554      The exact size is 3 * (inlen + (ctx ? ctx->i : 0)) / 4, minus 1 if the
 555      input ends with "=" and minus another 1 if the input ends with "==".
 556      Dividing before multiplying avoids the possibility of overflow.  */
 557   size_t needlen = 3 * (inlen / 4) + 3;
 558
 559   *out = malloc (needlen);
 560   if (!*out)
 561     return true;
 562
 563   if (!base64_decode_ctx (ctx, in, inlen, *out, &needlen))
 564     {
 565       free (*out);
 566       *out = NULL;
 567       return false;
 568     }
 569
 570   if (outlen)
 571     *outlen = needlen;
 572
 573   return true;
 574 }