windows/winutils.c

   1 /*
   2  * winutils.c: miscellaneous Windows utilities for GUI apps
   3  */
   4
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7 #include <ctype.h>
   8
   9 #include "misc.h"
  10
  11 #ifdef TESTMODE
  12 /* Definitions to allow this module to be compiled standalone for testing. */
  13 #define smalloc malloc
  14 #define srealloc realloc
  15 #define sfree free
  16 #endif
  17
  18 /*
  19  * Split a complete command line into argc/argv, attempting to do
  20  * it exactly the same way Windows itself would do it (so that
  21  * console utilities, which receive argc and argv from Windows,
  22  * will have their command lines processed in the same way as GUI
  23  * utilities which get a whole command line and must break it
  24  * themselves).
  25  *
  26  * Does not modify the input command line.
  27  *
  28  * The final parameter (argstart) is used to return a second array
  29  * of char * pointers, the same length as argv, each one pointing
  30  * at the start of the corresponding element of argv in the
  31  * original command line. So if you get half way through processing
  32  * your command line in argc/argv form and then decide you want to
  33  * treat the rest as a raw string, you can. If you don't want to,
  34  * `argstart' can be safely left NULL.
  35  */
  36 void split_into_argv(char *cmdline, int *argc, char ***argv,
  37                      char ***argstart)
  38 {
  39     char *p;
  40     char *outputline, *q;
  41     char **outputargv, **outputargstart;
  42     int outputargc;
  43
  44     /*
  45      * At first glance the rules appeared to be:
  46      *
  47      *  - Single quotes are not special characters.
  48      *
  49      *  - Double quotes are removed, but within them spaces cease
  50      *    to be special.
  51      *
  52      *  - Backslashes are _only_ special when a sequence of them
  53      *    appear just before a double quote. In this situation,
  54      *    they are treated like C backslashes: so \" just gives a
  55      *    literal quote, \\" gives a literal backslash and then
  56      *    opens or closes a double-quoted segment, \\\" gives a
  57      *    literal backslash and then a literal quote, \\\\" gives
  58      *    two literal backslashes and then opens/closes a
  59      *    double-quoted segment, and so forth. Note that this
  60      *    behaviour is identical inside and outside double quotes.
  61      *
  62      *  - Two successive double quotes become one literal double
  63      *    quote, but only _inside_ a double-quoted segment.
  64      *    Outside, they just form an empty double-quoted segment
  65      *    (which may cause an empty argument word).
  66      *
  67      *  - That only leaves the interesting question of what happens
  68      *    when one or more backslashes precedes two or more double
  69      *    quotes, starting inside a double-quoted string. And the
  70      *    answer to that appears somewhat bizarre. Here I tabulate
  71      *    number of backslashes (across the top) against number of
  72      *    quotes (down the left), and indicate how many backslashes
  73      *    are output, how many quotes are output, and whether a
  74      *    quoted segment is open at the end of the sequence:
  75      *
  76      *                      backslashes
  77      *
  78      *               0         1      2      3      4
  79      *
  80      *         0   0,0,y  |  1,0,y  2,0,y  3,0,y  4,0,y
  81      *            --------+-----------------------------
  82      *         1   0,0,n  |  0,1,y  1,0,n  1,1,y  2,0,n
  83      *    q    2   0,1,n  |  0,1,n  1,1,n  1,1,n  2,1,n
  84      *    u    3   0,1,y  |  0,2,n  1,1,y  1,2,n  2,1,y
  85      *    o    4   0,1,n  |  0,2,y  1,1,n  1,2,y  2,1,n
  86      *    t    5   0,2,n  |  0,2,n  1,2,n  1,2,n  2,2,n
  87      *    e    6   0,2,y  |  0,3,n  1,2,y  1,3,n  2,2,y
  88      *    s    7   0,2,n  |  0,3,y  1,2,n  1,3,y  2,2,n
  89      *         8   0,3,n  |  0,3,n  1,3,n  1,3,n  2,3,n
  90      *         9   0,3,y  |  0,4,n  1,3,y  1,4,n  2,3,y
  91      *        10   0,3,n  |  0,4,y  1,3,n  1,4,y  2,3,n
  92      *        11   0,4,n  |  0,4,n  1,4,n  1,4,n  2,4,n
  93      *
  94      *
  95      *      [Test fragment was of the form "a\\\"""b c" d.]
  96      *
  97      * There is very weird mod-3 behaviour going on here in the
  98      * number of quotes, and it even applies when there aren't any
  99      * backslashes! How ghastly.
 100      *
 101      * With a bit of thought, this extremely odd diagram suddenly
 102      * coalesced itself into a coherent, if still ghastly, model of
 103      * how things work:
 104      *
 105      *  - As before, backslashes are only special when one or more
 106      *    of them appear contiguously before at least one double
 107      *    quote. In this situation the backslashes do exactly what
 108      *    you'd expect: each one quotes the next thing in front of
 109      *    it, so you end up with n/2 literal backslashes (if n is
 110      *    even) or (n-1)/2 literal backslashes and a literal quote
 111      *    (if n is odd). In the latter case the double quote
 112      *    character right after the backslashes is used up.
 113      *
 114      *  - After that, any remaining double quotes are processed. A
 115      *    string of contiguous unescaped double quotes has a mod-3
 116      *    behaviour:
 117      *
 118      *     * inside a quoted segment, a quote ends the segment.
 119      *     * _immediately_ after ending a quoted segment, a quote
 120      *       simply produces a literal quote.
 121      *     * otherwise, outside a quoted segment, a quote begins a
 122      *       quoted segment.
 123      *
 124      *    So, for example, if we started inside a quoted segment
 125      *    then two contiguous quotes would close the segment and
 126      *    produce a literal quote; three would close the segment,
 127      *    produce a literal quote, and open a new segment. If we
 128      *    started outside a quoted segment, then two contiguous
 129      *    quotes would open and then close a segment, producing no
 130      *    output (but potentially creating a zero-length argument);
 131      *    but three quotes would open and close a segment and then
 132      *    produce a literal quote.
 133      */
 134
 135     /*
 136      * First deal with the simplest of all special cases: if there
 137      * aren't any arguments, return 0,NULL,NULL.
 138      */
 139     while (*cmdline && isspace(*cmdline)) cmdline++;
 140     if (!*cmdline) {
 141         if (argc) *argc = 0;
 142         if (argv) *argv = NULL;
 143         if (argstart) *argstart = NULL;
 144         return;
 145     }
 146
 147     /*
 148      * This will guaranteeably be big enough; we can realloc it
 149      * down later.
 150      */
 151     outputline = snewn(1+strlen(cmdline), char);
 152     outputargv = snewn(strlen(cmdline)+1 / 2, char *);
 153     outputargstart = snewn(strlen(cmdline)+1 / 2, char *);
 154
 155     p = cmdline; q = outputline; outputargc = 0;
 156
 157     while (*p) {
 158         int quote;
 159
 160         /* Skip whitespace searching for start of argument. */
 161         while (*p && isspace(*p)) p++;
 162         if (!*p) break;
 163
 164         /* We have an argument; start it. */
 165         outputargv[outputargc] = q;
 166         outputargstart[outputargc] = p;
 167         outputargc++;
 168         quote = 0;
 169
 170         /* Copy data into the argument until it's finished. */
 171         while (*p) {
 172             if (!quote && isspace(*p))
 173                 break;                 /* argument is finished */
 174
 175             if (*p == '"' || *p == '\\') {
 176                 /*
 177                  * We have a sequence of zero or more backslashes
 178                  * followed by a sequence of zero or more quotes.
 179                  * Count up how many of each, and then deal with
 180                  * them as appropriate.
 181                  */
 182                 int i, slashes = 0, quotes = 0;
 183                 while (*p == '\\') slashes++, p++;
 184                 while (*p == '"') quotes++, p++;
 185
 186                 if (!quotes) {
 187                     /*
 188                      * Special case: if there are no quotes,
 189                      * slashes are not special at all, so just copy
 190                      * n slashes to the output string.
 191                      */
 192                     while (slashes--) *q++ = '\\';
 193                 } else {
 194                     /* Slashes annihilate in pairs. */
 195                     while (slashes >= 2) slashes -= 2, *q++ = '\\';
 196
 197                     /* One remaining slash takes out the first quote. */
 198                     if (slashes) quotes--, *q++ = '"';
 199
 200                     if (quotes > 0) {
 201                         /* Outside a quote segment, a quote starts one. */
 202                         if (!quote) quotes--, quote = 1;
 203
 204                         /* Now we produce (n+1)/3 literal quotes... */
 205                         for (i = 3; i <= quotes+1; i += 3) *q++ = '"';
 206
 207                         /* ... and end in a quote segment iff 3 divides n. */
 208                         quote = (quotes % 3 == 0);
 209                     }
 210                 }
 211             } else {
 212                 *q++ = *p++;
 213             }
 214         }
 215
 216         /* At the end of an argument, just append a trailing NUL. */
 217         *q++ = '\0';
 218     }
 219
 220     outputargv = sresize(outputargv, outputargc, char *);
 221     outputargstart = sresize(outputargstart, outputargc, char *);
 222
 223     if (argc) *argc = outputargc;
 224     if (argv) *argv = outputargv; else sfree(outputargv);
 225     if (argstart) *argstart = outputargstart; else sfree(outputargstart);
 226 }
 227
 228 #ifdef TESTMODE
 229
 230 const struct argv_test {
 231     const char *cmdline;
 232     const char *argv[10];
 233 } argv_tests[] = {
 234     /*
 235      * We generate this set of tests by invoking ourself with
 236      * `-generate'.
 237      */
 238     {"ab c\" d", {"ab", "c d", NULL}},
 239     {"a\"b c\" d", {"ab c", "d", NULL}},
 240     {"a\"\"b c\" d", {"ab", "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     {"a\\\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
 255     {"a\\\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b 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     {"\"ab c\" d", {"ab c", "d", NULL}},
 284     {"\"a\"b c\" d", {"ab", "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     {"\"a\\\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b c", "d", NULL}},
 300     {"\"a\\\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
 301     {"\"a\\\\b c\" d", {"a\\\\b c", "d", NULL}},
 302     {"\"a\\\\\"b c\" d", {"a\\b", "c d", NULL}},
 303     {"\"a\\\\\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 304     {"\"a\\\\\"\"\"b c\" d", {"a\\\"b c", "d", NULL}},
 305     {"\"a\\\\\"\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 306     {"\"a\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 307     {"\"a\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
 308     {"\"a\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 309     {"\"a\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
 310     {"\"a\\\\\\b c\" d", {"a\\\\\\b c", "d", NULL}},
 311     {"\"a\\\\\\\"b c\" d", {"a\\\"b c", "d", NULL}},
 312     {"\"a\\\\\\\"\"b c\" d", {"a\\\"b", "c d", NULL}},
 313     {"\"a\\\\\\\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 314     {"\"a\\\\\\\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
 315     {"\"a\\\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
 316     {"\"a\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
 317     {"\"a\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b c", "d", NULL}},
 318     {"\"a\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
 319     {"\"a\\\\\\\\b c\" d", {"a\\\\\\\\b c", "d", NULL}},
 320     {"\"a\\\\\\\\\"b c\" d", {"a\\\\b", "c d", NULL}},
 321     {"\"a\\\\\\\\\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
 322     {"\"a\\\\\\\\\"\"\"b c\" d", {"a\\\\\"b c", "d", NULL}},
 323     {"\"a\\\\\\\\\"\"\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
 324     {"\"a\\\\\\\\\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
 325     {"\"a\\\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b c", "d", NULL}},
 326     {"\"a\\\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
 327     {"\"a\\\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"\"b", "c d", NULL}},
 328 };
 329
 330 int main(int argc, char **argv)
 331 {
 332     int i, j;
 333
 334     if (argc > 1) {
 335         /*
 336          * Generation of tests.
 337          *
 338          * Given `-splat <args>', we print out a C-style
 339          * representation of each argument (in the form "a", "b",
 340          * NULL), backslash-escaping each backslash and double
 341          * quote.
 342          *
 343          * Given `-split <string>', we first doctor `string' by
 344          * turning forward slashes into backslashes, single quotes
 345          * into double quotes and underscores into spaces; and then
 346          * we feed the resulting string to ourself with `-splat'.
 347          *
 348          * Given `-generate', we concoct a variety of fun test
 349          * cases, encode them in quote-safe form (mapping \, " and
 350          * space to /, ' and _ respectively) and feed each one to
 351          * `-split'.
 352          */
 353         if (!strcmp(argv[1], "-splat")) {
 354             int i;
 355             char *p;
 356             for (i = 2; i < argc; i++) {
 357                 putchar('"');
 358                 for (p = argv[i]; *p; p++) {
 359                     if (*p == '\\' || *p == '"')
 360                         putchar('\\');
 361                     putchar(*p);
 362                 }
 363                 printf("\", ");
 364             }
 365             printf("NULL");
 366             return 0;
 367         }
 368
 369         if (!strcmp(argv[1], "-split") && argc > 2) {
 370             char *str = malloc(20 + strlen(argv[0]) + strlen(argv[2]));
 371             char *p, *q;
 372
 373             q = str + sprintf(str, "%s -splat ", argv[0]);
 374             printf("    {\"");
 375             for (p = argv[2]; *p; p++, q++) {
 376                 switch (*p) {
 377                   case '/':  printf("\\\\"); *q = '\\'; break;
 378                   case '\'': printf("\\\""); *q = '"';  break;
 379                   case '_':  printf(" ");    *q = ' ';  break;
 380                   default:   putchar(*p);    *q = *p;   break;
 381                 }
 382             }
 383             *p = '\0';
 384             printf("\", {");
 385             fflush(stdout);
 386
 387             system(str);
 388
 389             printf("}},\n");
 390
 391             return 0;
 392         }
 393
 394         if (!strcmp(argv[1], "-generate")) {
 395             char *teststr, *p;
 396             int i, initialquote, backslashes, quotes;
 397
 398             teststr = malloc(200 + strlen(argv[0]));
 399
 400             for (initialquote = 0; initialquote <= 1; initialquote++) {
 401                 for (backslashes = 0; backslashes < 5; backslashes++) {
 402                     for (quotes = 0; quotes < 9; quotes++) {
 403                         p = teststr + sprintf(teststr, "%s -split ", argv[0]);
 404                         if (initialquote) *p++ = '\'';
 405                         *p++ = 'a';
 406                         for (i = 0; i < backslashes; i++) *p++ = '/';
 407                         for (i = 0; i < quotes; i++) *p++ = '\'';
 408                         *p++ = 'b';
 409                         *p++ = '_';
 410                         *p++ = 'c';
 411                         *p++ = '\'';
 412                         *p++ = '_';
 413                         *p++ = 'd';
 414                         *p = '\0';
 415
 416                         system(teststr);
 417                     }
 418                 }
 419             }
 420             return 0;
 421         }
 422
 423         fprintf(stderr, "unrecognised option: \"%s\"\n", argv[1]);
 424         return 1;
 425     }
 426
 427     /*
 428      * If we get here, we were invoked with no arguments, so just
 429      * run the tests.
 430      */
 431
 432     for (i = 0; i < lenof(argv_tests); i++) {
 433         int ac;
 434         char **av;
 435
 436         split_into_argv(argv_tests[i].cmdline, &ac, &av);
 437
 438         for (j = 0; j < ac && argv_tests[i].argv[j]; j++) {
 439             if (strcmp(av[j], argv_tests[i].argv[j])) {
 440                 printf("failed test %d (|%s|) arg %d: |%s| should be |%s|\n",
 441                        i, argv_tests[i].cmdline,
 442                        j, av[j], argv_tests[i].argv[j]);
 443             }
 444 #ifdef VERBOSE
 445             else {
 446                 printf("test %d (|%s|) arg %d: |%s| == |%s|\n",
 447                        i, argv_tests[i].cmdline,
 448                        j, av[j], argv_tests[i].argv[j]);
 449             }
 450 #endif
 451         }
 452         if (j < ac)
 453             printf("failed test %d (|%s|): %d args returned, should be %d\n",
 454                    i, argv_tests[i].cmdline, ac, j);
 455         if (argv_tests[i].argv[j])
 456             printf("failed test %d (|%s|): %d args returned, should be more\n",
 457                    i, argv_tests[i].cmdline, ac);
 458     }
 459
 460     return 0;
 461 }
 462
 463 #endif