gl/base64.c

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