winutils.c

   1 /*
   2  * winutils.c: miscellaneous Windows utilities
   3  */
   4
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7
   8 #define lenof(x) ( sizeof((x)) / sizeof(*(x)) )
   9
  10 #ifdef TESTMODE
  11 /* Definitions to allow this module to be compiled standalone for testing. */
  12 #define smalloc malloc
  13 #endif
  14
  15 /*
  16  * Split a complete command line into argc/argv, attempting to do
  17  * it exactly the same way Windows itself would do it (so that
  18  * console utilities, which receive argc and argv from Windows,
  19  * will have their command lines processed in the same way as GUI
  20  * utilities which get a whole command line and must break it
  21  * themselves).
  22  *
  23  * Does not modify the input command line (just in case).
  24  */
  25 void split_into_argv(const char *cmdline, int *argc, char ***argv)
  26 {
  27     const char *p;
  28     char *outputline, *q;
  29     char **outputargv;
  30     int outputargc;
  31
  32     /*
  33      * At first glance the rules appeared to be:
  34      *
  35      *  - Single quotes are not special characters.
  36      *
  37      *  - Double quotes are removed, but within them spaces cease
  38      *    to be special.
  39      *
  40      *  - Backslashes are _only_ special when a sequence of them
  41      *    appear just before a double quote. In this situation,
  42      *    they are treated like C backslashes: so \" just gives a
  43      *    literal quote, \\" gives a literal backslash and then
  44      *    opens or closes a double-quoted segment, \\\" gives a
  45      *    literal backslash and then a literal quote, \\\\" gives
  46      *    two literal backslashes and then opens/closes a
  47      *    double-quoted segment, and so forth. Note that this
  48      *    behaviour is identical inside and outside double quotes.
  49      *
  50      *  - Two successive double quotes become one literal double
  51      *    quote, but only _inside_ a double-quoted segment.
  52      *    Outside, they just form an empty double-quoted segment
  53      *    (which may cause an empty argument word).
  54      *
  55      *  - That only leaves the interesting question of what happens
  56      *    when one or more backslashes precedes two or more double
  57      *    quotes, starting inside a double-quoted string. And the
  58      *    answer to that appears somewhat bizarre. Here I tabulate
  59      *    number of backslashes (across the top) against number of
  60      *    quotes (down the left), and indicate how many backslashes
  61      *    are output, how many quotes are output, and whether a
  62      *    quoted segment is open at the end of the sequence:
  63      *
  64      *                      backslashes
  65      *
  66      *               0         1      2      3      4
  67      *
  68      *         0   0,0,y  |  1,0,y  2,0,y  3,0,y  4,0,y
  69      *            --------+-----------------------------
  70      *         1   0,0,n  |  0,1,y  1,0,n  1,1,y  2,0,n
  71      *    q    2   0,1,n  |  0,1,n  1,1,n  1,1,n  2,1,n
  72      *    u    3   0,1,y  |  0,2,n  1,1,y  1,2,n  2,1,y
  73      *    o    4   0,1,n  |  0,2,y  1,1,n  1,2,y  2,1,n
  74      *    t    5   0,2,n  |  0,2,n  1,2,n  1,2,n  2,2,n
  75      *    e    6   0,2,y  |  0,3,n  1,2,y  1,3,n  2,2,y
  76      *    s    7   0,2,n  |  0,3,y  1,2,n  1,3,y  2,2,n
  77      *         8   0,3,n  |  0,3,n  1,3,n  1,3,n  2,3,n
  78      *         9   0,3,y  |  0,4,n  1,3,y  1,4,n  2,3,y
  79      *        10   0,3,n  |  0,4,y  1,3,n  1,4,y  2,3,n
  80      *        11   0,4,n  |  0,4,n  1,4,n  1,4,n  2,4,n
  81      *
  82      *
  83      *      [Test fragment was of the form "a\\\"""b c" d.]
  84      *
  85      * There is very weird mod-3 behaviour going on here in the
  86      * number of quotes, and it even applies when there aren't any
  87      * backslashes! How ghastly.
  88      *
  89      * With a bit of thought, this extremely odd diagram suddenly
  90      * coalesced itself into a coherent, if still ghastly, model of
  91      * how things work:
  92      *
  93      *  - As before, backslashes are only special when one or more
  94      *    of them appear contiguously before at least one double
  95      *    quote. In this situation the backslashes do exactly what
  96      *    you'd expect: each one quotes the next thing in front of
  97      *    it, so you end up with n/2 literal backslashes (if n is
  98      *    even) or (n-1)/2 literal backslashes and a literal quote
  99      *    (if n is odd). In the latter case the double quote
 100      *    character right after the backslashes is used up.
 101      *
 102      *  - After that, any remaining double quotes are processed. A
 103      *    string of contiguous unescaped double quotes has a mod-3
 104      *    behaviour:
 105      *
 106      *     * inside a quoted segment, a quote ends the segment.
 107      *     * _immediately_ after ending a quoted segment, a quote
 108      *       simply produces a literal quote.
 109      *     * otherwise, outside a quoted segment, a quote begins a
 110      *       quoted segment.
 111      *
 112      *    So, for example, if we started inside a quoted segment
 113      *    then two contiguous quotes would close the segment and
 114      *    produce a literal quote; three would close the segment,
 115      *    produce a literal quote, and open a new segment. If we
 116      *    started outside a quoted segment, then two contiguous
 117      *    quotes would open and then close a segment, producing no
 118      *    output (but potentially creating a zero-length argument);
 119      *    but three quotes would open and close a segment and then
 120      *    produce a literal quote.
 121      */
 122
 123     /*
 124      * This will guaranteeably be big enough; we can realloc it
 125      * down later.
 126      */
 127     outputline = malloc(1+strlen(cmdline));
 128     outputargv = malloc(sizeof(char *) * (strlen(cmdline)+1 / 2));
 129
 130     p = cmdline; q = outputline; outputargc = 0;
 131
 132     while (*p) {
 133         int quote;
 134
 135         /* Skip whitespace searching for start of argument. */
 136         while (*p && isspace(*p)) p++;
 137         if (!*p) break;
 138
 139         /* We have an argument; start it. */
 140         outputargv[outputargc++] = q;
 141         quote = 0;
 142
 143         /* Copy data into the argument until it's finished. */
 144         while (*p) {
 145             if (!quote && isspace(*p))
 146                 break;                 /* argument is finished */
 147
 148             if (*p == '"' || *p == '\\') {
 149                 /*
 150                  * We have a sequence of zero or more backslashes
 151                  * followed by a sequence of zero or more quotes.
 152                  * Count up how many of each, and then deal with
 153                  * them as appropriate.
 154                  */
 155                 int i, slashes = 0, quotes = 0;
 156                 while (*p == '\\') slashes++, p++;
 157                 while (*p == '"') quotes++, p++;
 158
 159                 if (!quotes) {
 160                     /*
 161                      * Special case: if there are no quotes,
 162                      * slashes are not special at all, so just copy
 163                      * n slashes to the output string.
 164                      */
 165                     while (slashes--) *q++ = '\\';
 166                 } else {
 167                     /* Slashes annihilate in pairs. */
 168                     while (slashes >= 2) slashes -= 2, *q++ = '\\';
 169
 170                     /* One remaining slash takes out the first quote. */
 171                     if (slashes) quotes--, *q++ = '"';
 172
 173                     if (quotes > 0) {
 174                         /* Outside a quote segment, a quote starts one. */
 175                         if (!quote) quotes--, quote = 1;
 176
 177                         /* Now we produce (n+1)/3 literal quotes... */
 178                         for (i = 3; i <= quotes+1; i += 3) *q++ = '"';
 179
 180                         /* ... and end in a quote segment iff 3 divides n. */
 181                         quote = (quotes % 3 == 0);
 182                     }
 183                 }
 184             } else {
 185                 *q++ = *p++;
 186             }
 187         }
 188
 189         /* At the end of an argument, just append a trailing NUL. */
 190         *q++ = '\0';
 191     }
 192
 193     outputargv = realloc(outputargv, sizeof(char *) * outputargc);
 194
 195     if (argc) *argc = outputargc;
 196     if (argv) *argv = outputargv;
 197 }
 198
 199 #ifdef TESTMODE
 200
 201 const struct argv_test {
 202     const char *cmdline;
 203     const char *argv[10];
 204 } argv_tests[] = {
 205     /*
 206      * We generate this set of tests by invoking ourself with
 207      * `-generate'.
 208      */
 209     {"ab c\" d", {"ab", "c d", NULL}},
 210     {"a\"b c\" d", {"ab c", "d", NULL}},
 211     {"a\"\"b c\" d", {"ab", "c d", NULL}},
 212     {"a\"\"\"b c\" d", {"a\"b", "c d", NULL}},
 213     {"a\"\"\"\"b c\" d", {"a\"b c", "d", NULL}},
 214     {"a\"\"\"\"\"b c\" d", {"a\"b", "c d", NULL}},
 215     {"a\"\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 216     {"a\"\"\"\"\"\"\"b c\" d", {"a\"\"b c", "d", NULL}},
 217     {"a\"\"\"\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 218     {"a\\b c\" d", {"a\\b", "c d", NULL}},
 219     {"a\\\"b c\" d", {"a\"b", "c d", NULL}},
 220     {"a\\\"\"b c\" d", {"a\"b c", "d", NULL}},
 221     {"a\\\"\"\"b c\" d", {"a\"b", "c d", NULL}},
 222     {"a\\\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 223     {"a\\\"\"\"\"\"b c\" d", {"a\"\"b c", "d", NULL}},
 224     {"a\\\"\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 225     {"a\\\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
 226     {"a\\\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b c", "d", NULL}},
 227     {"a\\\\b c\" d", {"a\\\\b", "c d", NULL}},
 228     {"a\\\\\"b c\" d", {"a\\b c", "d", NULL}},
 229     {"a\\\\\"\"b c\" d", {"a\\b", "c d", NULL}},
 230     {"a\\\\\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 231     {"a\\\\\"\"\"\"b c\" d", {"a\\\"b c", "d", NULL}},
 232     {"a\\\\\"\"\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 233     {"a\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 234     {"a\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
 235     {"a\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 236     {"a\\\\\\b c\" d", {"a\\\\\\b", "c d", NULL}},
 237     {"a\\\\\\\"b c\" d", {"a\\\"b", "c d", NULL}},
 238     {"a\\\\\\\"\"b c\" d", {"a\\\"b c", "d", NULL}},
 239     {"a\\\\\\\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 240     {"a\\\\\\\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 241     {"a\\\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
 242     {"a\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 243     {"a\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
 244     {"a\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b c", "d", NULL}},
 245     {"a\\\\\\\\b c\" d", {"a\\\\\\\\b", "c d", NULL}},
 246     {"a\\\\\\\\\"b c\" d", {"a\\\\b c", "d", NULL}},
 247     {"a\\\\\\\\\"\"b c\" d", {"a\\\\b", "c d", NULL}},
 248     {"a\\\\\\\\\"\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
 249     {"a\\\\\\\\\"\"\"\"b c\" d", {"a\\\\\"b c", "d", NULL}},
 250     {"a\\\\\\\\\"\"\"\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
 251     {"a\\\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
 252     {"a\\\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b c", "d", NULL}},
 253     {"a\\\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
 254     {"\"ab c\" d", {"ab c", "d", NULL}},
 255     {"\"a\"b c\" d", {"ab", "c d", NULL}},
 256     {"\"a\"\"b c\" d", {"a\"b", "c d", NULL}},
 257     {"\"a\"\"\"b c\" d", {"a\"b c", "d", NULL}},
 258     {"\"a\"\"\"\"b c\" d", {"a\"b", "c d", NULL}},
 259     {"\"a\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 260     {"\"a\"\"\"\"\"\"b c\" d", {"a\"\"b c", "d", NULL}},
 261     {"\"a\"\"\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 262     {"\"a\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
 263     {"\"a\\b c\" d", {"a\\b c", "d", NULL}},
 264     {"\"a\\\"b c\" d", {"a\"b c", "d", NULL}},
 265     {"\"a\\\"\"b c\" d", {"a\"b", "c d", NULL}},
 266     {"\"a\\\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 267     {"\"a\\\"\"\"\"b c\" d", {"a\"\"b c", "d", NULL}},
 268     {"\"a\\\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
 269     {"\"a\\\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
 270     {"\"a\\\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b c", "d", NULL}},
 271     {"\"a\\\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
 272     {"\"a\\\\b c\" d", {"a\\\\b c", "d", NULL}},
 273     {"\"a\\\\\"b c\" d", {"a\\b", "c d", NULL}},
 274     {"\"a\\\\\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 275     {"\"a\\\\\"\"\"b c\" d", {"a\\\"b c", "d", NULL}},
 276     {"\"a\\\\\"\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 277     {"\"a\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 278     {"\"a\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
 279     {"\"a\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 280     {"\"a\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
 281     {"\"a\\\\\\b c\" d", {"a\\\\\\b c", "d", NULL}},
 282     {"\"a\\\\\\\"b c\" d", {"a\\\"b c", "d", NULL}},
 283     {"\"a\\\\\\\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 284     {"\"a\\\\\\\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 285     {"\"a\\\\\\\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
 286     {"\"a\\\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 287     {"\"a\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
 288     {"\"a\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b c", "d", NULL}},
 289     {"\"a\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
 290     {"\"a\\\\\\\\b c\" d", {"a\\\\\\\\b c", "d", NULL}},
 291     {"\"a\\\\\\\\\"b c\" d", {"a\\\\b", "c d", NULL}},
 292     {"\"a\\\\\\\\\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
 293     {"\"a\\\\\\\\\"\"\"b c\" d", {"a\\\\\"b c", "d", NULL}},
 294     {"\"a\\\\\\\\\"\"\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
 295     {"\"a\\\\\\\\\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
 296     {"\"a\\\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b c", "d", NULL}},
 297     {"\"a\\\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
 298     {"\"a\\\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"\"b", "c d", NULL}},
 299 };
 300
 301 int main(int argc, char **argv)
 302 {
 303     int i, j;
 304
 305     if (argc > 1) {
 306         /*
 307          * Generation of tests.
 308          *
 309          * Given `-splat <args>', we print out a C-style
 310          * representation of each argument (in the form "a", "b",
 311          * NULL), backslash-escaping each backslash and double
 312          * quote.
 313          *
 314          * Given `-split <string>', we first doctor `string' by
 315          * turning forward slashes into backslashes, single quotes
 316          * into double quotes and underscores into spaces; and then
 317          * we feed the resulting string to ourself with `-splat'.
 318          *
 319          * Given `-generate', we concoct a variety of fun test
 320          * cases, encode them in quote-safe form (mapping \, " and
 321          * space to /, ' and _ respectively) and feed each one to
 322          * `-split'.
 323          */
 324         if (!strcmp(argv[1], "-splat")) {
 325             int i;
 326             char *p;
 327             for (i = 2; i < argc; i++) {
 328                 putchar('"');
 329                 for (p = argv[i]; *p; p++) {
 330                     if (*p == '\\' || *p == '"')
 331                         putchar('\\');
 332                     putchar(*p);
 333                 }
 334                 printf("\", ");
 335             }
 336             printf("NULL");
 337             return 0;
 338         }
 339
 340         if (!strcmp(argv[1], "-split") && argc > 2) {
 341             char *str = malloc(20 + strlen(argv[0]) + strlen(argv[2]));
 342             char *p, *q;
 343
 344             q = str + sprintf(str, "%s -splat ", argv[0]);
 345             printf("    {\"");
 346             for (p = argv[2]; *p; p++, q++) {
 347                 switch (*p) {
 348                   case '/':  printf("\\\\"); *q = '\\'; break;
 349                   case '\'': printf("\\\""); *q = '"';  break;
 350                   case '_':  printf(" ");    *q = ' ';  break;
 351                   default:   putchar(*p);    *q = *p;   break;
 352                 }
 353             }
 354             *p = '\0';
 355             printf("\", {");
 356             fflush(stdout);
 357
 358             system(str);
 359
 360             printf("}},\n");
 361
 362             return 0;
 363         }
 364
 365         if (!strcmp(argv[1], "-generate")) {
 366             char *teststr, *p;
 367             int i, initialquote, backslashes, quotes;
 368
 369             teststr = malloc(200 + strlen(argv[0]));
 370
 371             for (initialquote = 0; initialquote <= 1; initialquote++) {
 372                 for (backslashes = 0; backslashes < 5; backslashes++) {
 373                     for (quotes = 0; quotes < 9; quotes++) {
 374                         p = teststr + sprintf(teststr, "%s -split ", argv[0]);
 375                         if (initialquote) *p++ = '\'';
 376                         *p++ = 'a';
 377                         for (i = 0; i < backslashes; i++) *p++ = '/';
 378                         for (i = 0; i < quotes; i++) *p++ = '\'';
 379                         *p++ = 'b';
 380                         *p++ = '_';
 381                         *p++ = 'c';
 382                         *p++ = '\'';
 383                         *p++ = '_';
 384                         *p++ = 'd';
 385                         *p = '\0';
 386
 387                         system(teststr);
 388                     }
 389                 }
 390             }
 391             return 0;
 392         }
 393
 394         fprintf(stderr, "unrecognised option: \"%s\"\n", argv[1]);
 395         return 1;
 396     }
 397
 398     /*
 399      * If we get here, we were invoked with no arguments, so just
 400      * run the tests.
 401      */
 402
 403     for (i = 0; i < lenof(argv_tests); i++) {
 404         int ac;
 405         char **av;
 406
 407         split_into_argv(argv_tests[i].cmdline, &ac, &av);
 408
 409         for (j = 0; j < ac && argv_tests[i].argv[j]; j++) {
 410             if (strcmp(av[j], argv_tests[i].argv[j])) {
 411                 printf("failed test %d (|%s|) arg %d: |%s| should be |%s|\n",
 412                        i, argv_tests[i].cmdline,
 413                        j, av[j], argv_tests[i].argv[j]);
 414             }
 415 #ifdef VERBOSE
 416             else {
 417                 printf("test %d (|%s|) arg %d: |%s| == |%s|\n",
 418                        i, argv_tests[i].cmdline,
 419                        j, av[j], argv_tests[i].argv[j]);
 420             }
 421 #endif
 422         }
 423         if (j < ac)
 424             printf("failed test %d (|%s|): %d args returned, should be %d\n",
 425                    i, argv_tests[i].cmdline, ac, j);
 426         if (argv_tests[i].argv[j])
 427             printf("failed test %d (|%s|): %d args returned, should be more\n",
 428                    i, argv_tests[i].cmdline, ac);
 429     }
 430
 431     return 0;
 432 }
 433
 434 #endif