main.c

   1 /* base65536 en- and decoder with usage similar to GNU's base64.
   2 (C) 2016 Tobias Girstmair, http://isticktoit.net/
   3 Released under the GNU GPL v3. See LICENSE for details. */
   4
   5 #include <stdio.h>
   6 #include <string.h>
   7 #include <stdlib.h>
   8 #include "base65536.h"
   9
  10 #define VERSION "0.1"
  11
  12 #define DEFAULT_WRAP 72
  13 enum OPERATIONS {
  14         ENCODE,
  15         DECODE
  16 };
  17
  18 void do_encode (FILE* in, FILE* out, int wrap_column);
  19 void do_decode (FILE* in, FILE* out, int ignore_garbage);
  20
  21 int main (int argc, char** argv) {
  22         int operation = ENCODE;
  23         int wrap = DEFAULT_WRAP;
  24         int garb_ignore = 0;
  25         FILE* file_to_op = stdin;
  26
  27         for (int i = 1; i < argc; i++) {
  28         //TODO: not efficient and ugly
  29                 if (strcmp (argv[i], "-d") == 0 || strcmp (argv[i], "--decode") == 0) {
  30                         operation = DECODE;
  31                 } else if (strcmp (argv[i], "-i") == 0 || strcmp (argv[i], "--ignore-garbage") == 0) {
  32                         garb_ignore = 1;
  33                 } else if (strcmp (argv[i], "-w") == 0) {
  34                         if (i == argc-1) {
  35                                 fputs ("option required for argument -w", stderr);
  36                                 return EXIT_FAILURE;
  37                         } else {
  38                                 for (int j = 0; argv[i+1][j] != '\0'; j++)
  39                                         if (argv[i+1][j] < '0' || argv[i+1][j] > '9') {
  40                                                 fputs ("option required for argument -w", stderr);
  41                                                 return EXIT_FAILURE;
  42                                         }
  43
  44                                 wrap = atoi (argv[++i]);
  45                         }
  46                 } else if (strcmp (argv[i], "--help") == 0) {
  47                         fprintf (stderr, "help. ");
  48                         return EXIT_SUCCESS;
  49                 } else if (strcmp (argv[i], "--version") == 0) {
  50                         fprintf (stderr, "base65536 version %s", VERSION);
  51                         return EXIT_SUCCESS;
  52                 } else {
  53                         if (argv[i][0] == '-') {
  54                                 fprintf (stderr, "%s is not a valid argument.\n", argv[i]);
  55                                 return EXIT_FAILURE;
  56                         } else if (file_to_op == stdin) {
  57                                 file_to_op = fopen (argv[i], "rb");
  58                         } else {
  59                                 fputs ("cannot give more than one file.\n", stderr);
  60                                 return EXIT_FAILURE;
  61                         }
  62                 }
  63         }
  64
  65         if (operation == DECODE) {
  66                 do_decode (file_to_op, stdout, garb_ignore);
  67         } else {
  68                 do_encode (file_to_op, stdout, wrap);
  69         }
  70
  71         if (file_to_op != stdin) fclose (file_to_op);
  72
  73         return 0;
  74 }
  75
  76
  77 void do_encode (FILE* in, FILE* out, int wrap_column) {
  78         int in_char[2];
  79         char out_utf8[5];
  80         int unicode_cp;
  81         long cnt = 0;
  82
  83         while (1) {
  84                 if ((in_char[0] = getc(in)) == EOF) break;
  85                 in_char[1] = getc(in);
  86
  87                 unicode_cp = base65536_encode_char (in_char);
  88                 codepoint_to_utf8 (unicode_cp, out_utf8);
  89                 fprintf (out, "%s%s", wrap_column&&!(cnt%wrap_column)&&cnt?"\n":"", out_utf8);
  90                 cnt++;
  91                 if (in_char[1] == EOF) break;
  92         }
  93         printf ("\n");
  94 }
  95
  96 void do_decode (FILE* in, FILE* out, int ignore_garbage) {
  97         int out_char[2];
  98         char in_utf8[5];
  99         int unicode_cp;
 100
 101         while (1) {
 102                 for (int i = 0; i < 5; i++) {
 103                         int c = getc (in);
 104                         if (c == EOF) {
 105                                 return;
 106                         } else if (i != 0 && c < 0x80) { //start of new asciichar
 107                                 ungetc (c, in);
 108                                 in_utf8[i] = '\0';
 109                                 break;
 110                         } else if (i != 0 && c >= 0xc0) { //start of new utf8char
 111                                 ungetc (c, in);
 112                                 in_utf8[i] = '\0';
 113                                 break;
 114                         }
 115                         in_utf8[i] = c;
 116                 }
 117                 unicode_cp = utf8_to_codepoint (in_utf8);
 118                 //ignore ascii-chars, because base65k won't map to those (but
 119                 //whitespace may be added by the encoder / medium)
 120                 if (unicode_cp < 0x80) continue;
 121
 122                 if (base65536_decode_char (unicode_cp, out_char) > 0) {
 123                         fputc (out_char[0], out);
 124                         fputc (out_char[1], out);
 125                 } else {
 126                         if (!ignore_garbage) {
 127                                 fprintf (stderr, "Unrecognized glyph %s\n", in_utf8);
 128                                 return;
 129                         }
 130                 }
 131         }
 132 }