busybox/libbb/uuencode.c
<<
>>
Prefs
   1/* vi: set sw=4 ts=4: */
   2/*
   3 * Copyright 2003, Glenn McGrath
   4 * Copyright 2006, Rob Landley <rob@landley.net>
   5 * Copyright 2010, Denys Vlasenko
   6 *
   7 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
   8 */
   9#include "libbb.h"
  10
  11/* Conversion table.  for base 64 */
  12const char bb_uuenc_tbl_base64[65 + 1] ALIGN1 = {
  13        'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',
  14        'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P',
  15        'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
  16        'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f',
  17        'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n',
  18        'o', 'p', 'q', 'r', 's', 't', 'u', 'v',
  19        'w', 'x', 'y', 'z', '0', '1', '2', '3',
  20        '4', '5', '6', '7', '8', '9', '+', '/',
  21        '=' /* termination character */,
  22        '\0' /* needed for uudecode.c only */
  23};
  24
  25const char bb_uuenc_tbl_std[65] ALIGN1 = {
  26        '`', '!', '"', '#', '$', '%', '&', '\'',
  27        '(', ')', '*', '+', ',', '-', '.', '/',
  28        '0', '1', '2', '3', '4', '5', '6', '7',
  29        '8', '9', ':', ';', '<', '=', '>', '?',
  30        '@', 'A', 'B', 'C', 'D', 'E', 'F', 'G',
  31        'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O',
  32        'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',
  33        'X', 'Y', 'Z', '[', '\\', ']', '^', '_',
  34        '`' /* termination character */
  35};
  36
  37/*
  38 * Encode bytes at S of length LENGTH to uuencode or base64 format and place it
  39 * to STORE.  STORE will be 0-terminated, and must point to a writable
  40 * buffer of at least 1+BASE64_LENGTH(length) bytes.
  41 * where BASE64_LENGTH(len) = (4 * ((LENGTH + 2) / 3))
  42 */
  43void FAST_FUNC bb_uuencode(char *p, const void *src, int length, const char *tbl)
  44{
  45        const unsigned char *s = src;
  46
  47        /* Transform the 3x8 bits to 4x6 bits */
  48        while (length > 0) {
  49                unsigned s1, s2;
  50
  51                /* Are s[1], s[2] valid or should be assumed 0? */
  52                s1 = s2 = 0;
  53                length -= 3; /* can be >=0, -1, -2 */
  54                if (length >= -1) {
  55                        s1 = s[1];
  56                        if (length >= 0)
  57                                s2 = s[2];
  58                }
  59                *p++ = tbl[s[0] >> 2];
  60                *p++ = tbl[((s[0] & 3) << 4) + (s1 >> 4)];
  61                *p++ = tbl[((s1 & 0xf) << 2) + (s2 >> 6)];
  62                *p++ = tbl[s2 & 0x3f];
  63                s += 3;
  64        }
  65        /* Zero-terminate */
  66        *p = '\0';
  67        /* If length is -2 or -1, pad last char or two */
  68        while (length) {
  69                *--p = tbl[64];
  70                length++;
  71        }
  72}
  73
  74/*
  75 * Decode base64 encoded string. Stops on '\0'.
  76 *
  77 * Returns: pointer to the undecoded part of source.
  78 * If points to '\0', then the source was fully decoded.
  79 * (*pp_dst): advanced past the last written byte.
  80 */
  81const char* FAST_FUNC decode_base64(char **pp_dst, const char *src)
  82{
  83        char *dst = *pp_dst;
  84        const char *src_tail;
  85
  86        while (1) {
  87                unsigned char six_bit[4];
  88                int count = 0;
  89
  90                /* Fetch up to four 6-bit values */
  91                src_tail = src;
  92                while (count < 4) {
  93                        char *table_ptr;
  94                        int ch;
  95
  96                        /* Get next _valid_ character.
  97                         * bb_uuenc_tbl_base64[] contains this string:
  98                         *  0         1         2         3         4         5         6
  99                         *  01234567890123456789012345678901234567890123456789012345678901234
 100                         * "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/="
 101                         */
 102                        do {
 103                                ch = *src;
 104                                if (ch == '\0') {
 105                                        if (count == 0) {
 106                                                /* Example:
 107                                                 * If we decode "QUJD <NUL>", we want
 108                                                 * to return ptr to NUL, not to ' ',
 109                                                 * because we did fully decode
 110                                                 * the string (to "ABC").
 111                                                 */
 112                                                src_tail = src;
 113                                        }
 114                                        goto ret;
 115                                }
 116                                src++;
 117                                table_ptr = strchr(bb_uuenc_tbl_base64, ch);
 118//TODO: add BASE64_FLAG_foo to die on bad char?
 119                        } while (!table_ptr);
 120
 121                        /* Convert encoded character to decimal */
 122                        ch = table_ptr - bb_uuenc_tbl_base64;
 123
 124                        /* ch is 64 if char was '=', otherwise 0..63 */
 125                        if (ch == 64)
 126                                break;
 127                        six_bit[count] = ch;
 128                        count++;
 129                }
 130
 131                /* Transform 6-bit values to 8-bit ones.
 132                 * count can be < 4 when we decode the tail:
 133                 * "eQ==" -> "y", not "y NUL NUL".
 134                 * Note that (count > 1) is always true,
 135                 * "x===" encoding is not valid:
 136                 * even a single zero byte encodes as "AA==".
 137                 * However, with current logic we come here with count == 1
 138                 * when we decode "==" tail.
 139                 */
 140                if (count > 1)
 141                        *dst++ = six_bit[0] << 2 | six_bit[1] >> 4;
 142                if (count > 2)
 143                        *dst++ = six_bit[1] << 4 | six_bit[2] >> 2;
 144                if (count > 3)
 145                        *dst++ = six_bit[2] << 6 | six_bit[3];
 146                /* Note that if we decode "AA==" and ate first '=',
 147                 * we just decoded one char (count == 2) and now we'll
 148                 * do the loop once more to decode second '='.
 149                 */
 150        } /* while (1) */
 151 ret:
 152        *pp_dst = dst;
 153        return src_tail;
 154}
 155
 156/*
 157 * Decode base64 encoded stream.
 158 * Can stop on EOF, specified char, or on uuencode-style "====" line:
 159 * flags argument controls it.
 160 */
 161void FAST_FUNC read_base64(FILE *src_stream, FILE *dst_stream, int flags)
 162{
 163/* Note that EOF _can_ be passed as exit_char too */
 164#define exit_char    ((int)(signed char)flags)
 165#define uu_style_end (flags & BASE64_FLAG_UU_STOP)
 166
 167        /* uuencoded files have 61 byte lines. Use 64 byte buffer
 168         * to process line at a time.
 169         */
 170        enum { BUFFER_SIZE = 64 };
 171
 172        char in_buf[BUFFER_SIZE + 2];
 173        char out_buf[BUFFER_SIZE / 4 * 3 + 2];
 174        char *out_tail;
 175        const char *in_tail;
 176        int term_seen = 0;
 177        int in_count = 0;
 178
 179        while (1) {
 180                while (in_count < BUFFER_SIZE) {
 181                        int ch = fgetc(src_stream);
 182                        if (ch == exit_char) {
 183                                if (in_count == 0)
 184                                        return;
 185                                term_seen = 1;
 186                                break;
 187                        }
 188                        if (ch == EOF) {
 189                                term_seen = 1;
 190                                break;
 191                        }
 192                        /* Prevent "====" line to be split: stop if we see '\n'.
 193                         * We can also skip other whitespace and skirt the problem
 194                         * of files with NULs by stopping on any control char or space:
 195                         */
 196                        if (ch <= ' ')
 197                                break;
 198                        in_buf[in_count++] = ch;
 199                }
 200                in_buf[in_count] = '\0';
 201
 202                /* Did we encounter "====" line? */
 203                if (uu_style_end && strcmp(in_buf, "====") == 0)
 204                        return;
 205
 206                out_tail = out_buf;
 207                in_tail = decode_base64(&out_tail, in_buf);
 208
 209                fwrite(out_buf, (out_tail - out_buf), 1, dst_stream);
 210
 211                if (term_seen) {
 212                        /* Did we consume ALL characters? */
 213                        if (*in_tail == '\0')
 214                                return;
 215                        /* No */
 216                        bb_error_msg_and_die("truncated base64 input");
 217                }
 218
 219                /* It was partial decode */
 220                in_count = strlen(in_tail);
 221                memmove(in_buf, in_tail, in_count);
 222        }
 223}
 224