2 * winutils.c: miscellaneous Windows utilities for GUI apps
14 /* Definitions to allow this module to be compiled standalone for testing
15 * split_into_argv(). */
16 #define smalloc malloc
17 #define srealloc realloc
22 * GetOpenFileName/GetSaveFileName tend to muck around with the process'
23 * working directory on at least some versions of Windows.
24 * Here's a wrapper that gives more control over this, and hides a little
25 * bit of other grottiness.
33 * `of' is expected to be initialised with most interesting fields, but
34 * this function does some administrivia. (assume `of' was memset to 0)
35 * save==1 -> GetSaveFileName; save==0 -> GetOpenFileName
36 * `state' is optional.
38 BOOL request_file(filereq *state, OPENFILENAME *of, int preserve, int save)
40 TCHAR cwd[PATH_MAX]; /* process CWD */
45 DWORD r = GetCurrentDirectory(lenof(cwd), cwd);
46 if (r == 0 || r >= lenof(cwd))
47 /* Didn't work, oh well. Stop trying to be clever. */
51 /* Open the file requester, maybe setting lpstrInitialDir */
53 #ifdef OPENFILENAME_SIZE_VERSION_400
54 of->lStructSize = OPENFILENAME_SIZE_VERSION_400;
56 of->lStructSize = sizeof(*of);
58 of->lpstrInitialDir = (state && state->cwd[0]) ? state->cwd : NULL;
59 /* Actually put up the requester. */
60 ret = save ? GetSaveFileName(of) : GetOpenFileName(of);
63 /* Get CWD left by requester */
65 DWORD r = GetCurrentDirectory(lenof(state->cwd), state->cwd);
66 if (r == 0 || r >= lenof(state->cwd))
67 /* Didn't work, oh well. */
71 /* Restore process CWD */
73 /* If it fails, there's not much we can do. */
74 (void) SetCurrentDirectory(cwd);
79 filereq *filereq_new(void)
81 filereq *ret = snew(filereq);
86 void filereq_free(filereq *state)
92 * Split a complete command line into argc/argv, attempting to do
93 * it exactly the same way Windows itself would do it (so that
94 * console utilities, which receive argc and argv from Windows,
95 * will have their command lines processed in the same way as GUI
96 * utilities which get a whole command line and must break it
99 * Does not modify the input command line.
101 * The final parameter (argstart) is used to return a second array
102 * of char * pointers, the same length as argv, each one pointing
103 * at the start of the corresponding element of argv in the
104 * original command line. So if you get half way through processing
105 * your command line in argc/argv form and then decide you want to
106 * treat the rest as a raw string, you can. If you don't want to,
107 * `argstart' can be safely left NULL.
109 void split_into_argv(char *cmdline, int *argc, char ***argv,
113 char *outputline, *q;
114 char **outputargv, **outputargstart;
118 * At first glance the rules appeared to be:
120 * - Single quotes are not special characters.
122 * - Double quotes are removed, but within them spaces cease
125 * - Backslashes are _only_ special when a sequence of them
126 * appear just before a double quote. In this situation,
127 * they are treated like C backslashes: so \" just gives a
128 * literal quote, \\" gives a literal backslash and then
129 * opens or closes a double-quoted segment, \\\" gives a
130 * literal backslash and then a literal quote, \\\\" gives
131 * two literal backslashes and then opens/closes a
132 * double-quoted segment, and so forth. Note that this
133 * behaviour is identical inside and outside double quotes.
135 * - Two successive double quotes become one literal double
136 * quote, but only _inside_ a double-quoted segment.
137 * Outside, they just form an empty double-quoted segment
138 * (which may cause an empty argument word).
140 * - That only leaves the interesting question of what happens
141 * when one or more backslashes precedes two or more double
142 * quotes, starting inside a double-quoted string. And the
143 * answer to that appears somewhat bizarre. Here I tabulate
144 * number of backslashes (across the top) against number of
145 * quotes (down the left), and indicate how many backslashes
146 * are output, how many quotes are output, and whether a
147 * quoted segment is open at the end of the sequence:
153 * 0 0,0,y | 1,0,y 2,0,y 3,0,y 4,0,y
154 * --------+-----------------------------
155 * 1 0,0,n | 0,1,y 1,0,n 1,1,y 2,0,n
156 * q 2 0,1,n | 0,1,n 1,1,n 1,1,n 2,1,n
157 * u 3 0,1,y | 0,2,n 1,1,y 1,2,n 2,1,y
158 * o 4 0,1,n | 0,2,y 1,1,n 1,2,y 2,1,n
159 * t 5 0,2,n | 0,2,n 1,2,n 1,2,n 2,2,n
160 * e 6 0,2,y | 0,3,n 1,2,y 1,3,n 2,2,y
161 * s 7 0,2,n | 0,3,y 1,2,n 1,3,y 2,2,n
162 * 8 0,3,n | 0,3,n 1,3,n 1,3,n 2,3,n
163 * 9 0,3,y | 0,4,n 1,3,y 1,4,n 2,3,y
164 * 10 0,3,n | 0,4,y 1,3,n 1,4,y 2,3,n
165 * 11 0,4,n | 0,4,n 1,4,n 1,4,n 2,4,n
168 * [Test fragment was of the form "a\\\"""b c" d.]
170 * There is very weird mod-3 behaviour going on here in the
171 * number of quotes, and it even applies when there aren't any
172 * backslashes! How ghastly.
174 * With a bit of thought, this extremely odd diagram suddenly
175 * coalesced itself into a coherent, if still ghastly, model of
178 * - As before, backslashes are only special when one or more
179 * of them appear contiguously before at least one double
180 * quote. In this situation the backslashes do exactly what
181 * you'd expect: each one quotes the next thing in front of
182 * it, so you end up with n/2 literal backslashes (if n is
183 * even) or (n-1)/2 literal backslashes and a literal quote
184 * (if n is odd). In the latter case the double quote
185 * character right after the backslashes is used up.
187 * - After that, any remaining double quotes are processed. A
188 * string of contiguous unescaped double quotes has a mod-3
191 * * inside a quoted segment, a quote ends the segment.
192 * * _immediately_ after ending a quoted segment, a quote
193 * simply produces a literal quote.
194 * * otherwise, outside a quoted segment, a quote begins a
197 * So, for example, if we started inside a quoted segment
198 * then two contiguous quotes would close the segment and
199 * produce a literal quote; three would close the segment,
200 * produce a literal quote, and open a new segment. If we
201 * started outside a quoted segment, then two contiguous
202 * quotes would open and then close a segment, producing no
203 * output (but potentially creating a zero-length argument);
204 * but three quotes would open and close a segment and then
205 * produce a literal quote.
209 * First deal with the simplest of all special cases: if there
210 * aren't any arguments, return 0,NULL,NULL.
212 while (*cmdline && isspace(*cmdline)) cmdline++;
215 if (argv) *argv = NULL;
216 if (argstart) *argstart = NULL;
221 * This will guaranteeably be big enough; we can realloc it
224 outputline = snewn(1+strlen(cmdline), char);
225 outputargv = snewn(strlen(cmdline)+1 / 2, char *);
226 outputargstart = snewn(strlen(cmdline)+1 / 2, char *);
228 p = cmdline; q = outputline; outputargc = 0;
233 /* Skip whitespace searching for start of argument. */
234 while (*p && isspace(*p)) p++;
237 /* We have an argument; start it. */
238 outputargv[outputargc] = q;
239 outputargstart[outputargc] = p;
243 /* Copy data into the argument until it's finished. */
245 if (!quote && isspace(*p))
246 break; /* argument is finished */
248 if (*p == '"' || *p == '\\') {
250 * We have a sequence of zero or more backslashes
251 * followed by a sequence of zero or more quotes.
252 * Count up how many of each, and then deal with
253 * them as appropriate.
255 int i, slashes = 0, quotes = 0;
256 while (*p == '\\') slashes++, p++;
257 while (*p == '"') quotes++, p++;
261 * Special case: if there are no quotes,
262 * slashes are not special at all, so just copy
263 * n slashes to the output string.
265 while (slashes--) *q++ = '\\';
267 /* Slashes annihilate in pairs. */
268 while (slashes >= 2) slashes -= 2, *q++ = '\\';
270 /* One remaining slash takes out the first quote. */
271 if (slashes) quotes--, *q++ = '"';
274 /* Outside a quote segment, a quote starts one. */
275 if (!quote) quotes--, quote = 1;
277 /* Now we produce (n+1)/3 literal quotes... */
278 for (i = 3; i <= quotes+1; i += 3) *q++ = '"';
280 /* ... and end in a quote segment iff 3 divides n. */
281 quote = (quotes % 3 == 0);
289 /* At the end of an argument, just append a trailing NUL. */
293 outputargv = sresize(outputargv, outputargc, char *);
294 outputargstart = sresize(outputargstart, outputargc, char *);
296 if (argc) *argc = outputargc;
297 if (argv) *argv = outputargv; else sfree(outputargv);
298 if (argstart) *argstart = outputargstart; else sfree(outputargstart);
303 const struct argv_test {
305 const char *argv[10];
308 * We generate this set of tests by invoking ourself with
311 {"ab c\" d", {"ab", "c d", NULL}},
312 {"a\"b c\" d", {"ab c", "d", NULL}},
313 {"a\"\"b c\" d", {"ab", "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 {"a\\\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b c", "d", NULL}},
329 {"a\\\\b c\" d", {"a\\\\b", "c d", NULL}},
330 {"a\\\\\"b c\" d", {"a\\b c", "d", NULL}},
331 {"a\\\\\"\"b c\" d", {"a\\b", "c d", NULL}},
332 {"a\\\\\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
333 {"a\\\\\"\"\"\"b c\" d", {"a\\\"b c", "d", NULL}},
334 {"a\\\\\"\"\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
335 {"a\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
336 {"a\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
337 {"a\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
338 {"a\\\\\\b c\" d", {"a\\\\\\b", "c d", NULL}},
339 {"a\\\\\\\"b c\" d", {"a\\\"b", "c d", NULL}},
340 {"a\\\\\\\"\"b c\" d", {"a\\\"b c", "d", NULL}},
341 {"a\\\\\\\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
342 {"a\\\\\\\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
343 {"a\\\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
344 {"a\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
345 {"a\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
346 {"a\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b c", "d", NULL}},
347 {"a\\\\\\\\b c\" d", {"a\\\\\\\\b", "c d", NULL}},
348 {"a\\\\\\\\\"b c\" d", {"a\\\\b c", "d", NULL}},
349 {"a\\\\\\\\\"\"b c\" d", {"a\\\\b", "c d", NULL}},
350 {"a\\\\\\\\\"\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
351 {"a\\\\\\\\\"\"\"\"b c\" d", {"a\\\\\"b c", "d", NULL}},
352 {"a\\\\\\\\\"\"\"\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
353 {"a\\\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
354 {"a\\\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b c", "d", NULL}},
355 {"a\\\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
356 {"\"ab c\" d", {"ab c", "d", NULL}},
357 {"\"a\"b c\" d", {"ab", "c d", NULL}},
358 {"\"a\"\"b c\" d", {"a\"b", "c d", NULL}},
359 {"\"a\"\"\"b c\" d", {"a\"b c", "d", NULL}},
360 {"\"a\"\"\"\"b c\" d", {"a\"b", "c d", NULL}},
361 {"\"a\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
362 {"\"a\"\"\"\"\"\"b c\" d", {"a\"\"b c", "d", NULL}},
363 {"\"a\"\"\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
364 {"\"a\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
365 {"\"a\\b c\" d", {"a\\b c", "d", NULL}},
366 {"\"a\\\"b c\" d", {"a\"b c", "d", NULL}},
367 {"\"a\\\"\"b c\" d", {"a\"b", "c d", NULL}},
368 {"\"a\\\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
369 {"\"a\\\"\"\"\"b c\" d", {"a\"\"b c", "d", NULL}},
370 {"\"a\\\"\"\"\"\"b c\" d", {"a\"\"b", "c d", NULL}},
371 {"\"a\\\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
372 {"\"a\\\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b c", "d", NULL}},
373 {"\"a\\\"\"\"\"\"\"\"\"b c\" d", {"a\"\"\"b", "c d", NULL}},
374 {"\"a\\\\b c\" d", {"a\\\\b c", "d", NULL}},
375 {"\"a\\\\\"b c\" d", {"a\\b", "c d", NULL}},
376 {"\"a\\\\\"\"b c\" d", {"a\\\"b", "c d", NULL}},
377 {"\"a\\\\\"\"\"b c\" d", {"a\\\"b c", "d", NULL}},
378 {"\"a\\\\\"\"\"\"b c\" d", {"a\\\"b", "c d", NULL}},
379 {"\"a\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
380 {"\"a\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
381 {"\"a\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
382 {"\"a\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
383 {"\"a\\\\\\b c\" d", {"a\\\\\\b c", "d", NULL}},
384 {"\"a\\\\\\\"b c\" d", {"a\\\"b c", "d", NULL}},
385 {"\"a\\\\\\\"\"b c\" d", {"a\\\"b", "c d", NULL}},
386 {"\"a\\\\\\\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
387 {"\"a\\\\\\\"\"\"\"b c\" d", {"a\\\"\"b c", "d", NULL}},
388 {"\"a\\\\\\\"\"\"\"\"b c\" d", {"a\\\"\"b", "c d", NULL}},
389 {"\"a\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
390 {"\"a\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b c", "d", NULL}},
391 {"\"a\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\"\"\"b", "c d", NULL}},
392 {"\"a\\\\\\\\b c\" d", {"a\\\\\\\\b c", "d", NULL}},
393 {"\"a\\\\\\\\\"b c\" d", {"a\\\\b", "c d", NULL}},
394 {"\"a\\\\\\\\\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
395 {"\"a\\\\\\\\\"\"\"b c\" d", {"a\\\\\"b c", "d", NULL}},
396 {"\"a\\\\\\\\\"\"\"\"b c\" d", {"a\\\\\"b", "c d", NULL}},
397 {"\"a\\\\\\\\\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
398 {"\"a\\\\\\\\\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b c", "d", NULL}},
399 {"\"a\\\\\\\\\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"b", "c d", NULL}},
400 {"\"a\\\\\\\\\"\"\"\"\"\"\"\"b c\" d", {"a\\\\\"\"\"b", "c d", NULL}},
403 int main(int argc, char **argv)
409 * Generation of tests.
411 * Given `-splat <args>', we print out a C-style
412 * representation of each argument (in the form "a", "b",
413 * NULL), backslash-escaping each backslash and double
416 * Given `-split <string>', we first doctor `string' by
417 * turning forward slashes into backslashes, single quotes
418 * into double quotes and underscores into spaces; and then
419 * we feed the resulting string to ourself with `-splat'.
421 * Given `-generate', we concoct a variety of fun test
422 * cases, encode them in quote-safe form (mapping \, " and
423 * space to /, ' and _ respectively) and feed each one to
426 if (!strcmp(argv[1], "-splat")) {
429 for (i = 2; i < argc; i++) {
431 for (p = argv[i]; *p; p++) {
432 if (*p == '\\' || *p == '"')
442 if (!strcmp(argv[1], "-split") && argc > 2) {
443 char *str = malloc(20 + strlen(argv[0]) + strlen(argv[2]));
446 q = str + sprintf(str, "%s -splat ", argv[0]);
448 for (p = argv[2]; *p; p++, q++) {
450 case '/': printf("\\\\"); *q = '\\'; break;
451 case '\'': printf("\\\""); *q = '"'; break;
452 case '_': printf(" "); *q = ' '; break;
453 default: putchar(*p); *q = *p; break;
467 if (!strcmp(argv[1], "-generate")) {
469 int i, initialquote, backslashes, quotes;
471 teststr = malloc(200 + strlen(argv[0]));
473 for (initialquote = 0; initialquote <= 1; initialquote++) {
474 for (backslashes = 0; backslashes < 5; backslashes++) {
475 for (quotes = 0; quotes < 9; quotes++) {
476 p = teststr + sprintf(teststr, "%s -split ", argv[0]);
477 if (initialquote) *p++ = '\'';
479 for (i = 0; i < backslashes; i++) *p++ = '/';
480 for (i = 0; i < quotes; i++) *p++ = '\'';
496 fprintf(stderr, "unrecognised option: \"%s\"\n", argv[1]);
501 * If we get here, we were invoked with no arguments, so just
505 for (i = 0; i < lenof(argv_tests); i++) {
509 split_into_argv(argv_tests[i].cmdline, &ac, &av);
511 for (j = 0; j < ac && argv_tests[i].argv[j]; j++) {
512 if (strcmp(av[j], argv_tests[i].argv[j])) {
513 printf("failed test %d (|%s|) arg %d: |%s| should be |%s|\n",
514 i, argv_tests[i].cmdline,
515 j, av[j], argv_tests[i].argv[j]);
519 printf("test %d (|%s|) arg %d: |%s| == |%s|\n",
520 i, argv_tests[i].cmdline,
521 j, av[j], argv_tests[i].argv[j]);
526 printf("failed test %d (|%s|): %d args returned, should be %d\n",
527 i, argv_tests[i].cmdline, ac, j);
528 if (argv_tests[i].argv[j])
529 printf("failed test %d (|%s|): %d args returned, should be more\n",
530 i, argv_tests[i].cmdline, ac);