| 1 | ///////////////////////////////////////////////////////////////////////////// |
| 2 | // Name: glob.inc |
| 3 | // Purpose: File and other globbing (included by utils.cpp) |
| 4 | // Author: Karsten Ballueder |
| 5 | // Modified by: |
| 6 | // Created: 20/5/1998 |
| 7 | // RCS-ID: $Id$ |
| 8 | // Copyright: (c) Karsten Ballueder |
| 9 | // Licence: wxWindows licence |
| 10 | ///////////////////////////////////////////////////////////////////////////// |
| 11 | |
| 12 | // This file includes: |
| 13 | // wxIsWild(const char *pattern) |
| 14 | // wxMatchWild(const char *pattern, const char *str, bool dot_special) |
| 15 | // |
| 16 | |
| 17 | //--------------------------------------------------------------------------------- |
| 18 | #ifndef UNIX_GLOB |
| 19 | # ifdef wx_x |
| 20 | # define UNIX_GLOB 1 |
| 21 | # else |
| 22 | # define UNIX_GLOB 0 |
| 23 | # endif |
| 24 | #endif |
| 25 | |
| 26 | #if UNIX_GLOB |
| 27 | # ifdef wx_msw |
| 28 | # error "Can't use Unix file globbing under Windows!" |
| 29 | # endif |
| 30 | #endif |
| 31 | |
| 32 | /************************************************************************* |
| 33 | * |
| 34 | * wxIsWild checks whether the pattern contains wildcards, and |
| 35 | * returns TRUE if it does, and FALSE if it does not (or if the |
| 36 | * pattern is NULL -- i.e. no string). |
| 37 | * |
| 38 | * The argument is: |
| 39 | * |
| 40 | * 1) pattern - a character string |
| 41 | */ |
| 42 | bool |
| 43 | wxIsWild (const char *pattern) |
| 44 | { |
| 45 | while (*pattern) |
| 46 | { |
| 47 | switch (*pattern++) |
| 48 | { |
| 49 | case '?': |
| 50 | case '*': |
| 51 | #if UNIX_GLOB |
| 52 | case '[': |
| 53 | case '{': /* } */ |
| 54 | #endif |
| 55 | return TRUE; |
| 56 | #if UNIX_GLOB |
| 57 | case '\\': |
| 58 | if (!*pattern++) |
| 59 | return FALSE; |
| 60 | #endif |
| 61 | } /* switch() */ |
| 62 | } /* while() */ |
| 63 | return FALSE; |
| 64 | } |
| 65 | |
| 66 | |
| 67 | |
| 68 | #if UNIX_GLOB |
| 69 | |
| 70 | // Unix Glob() |
| 71 | // |
| 72 | // Pattern Function |
| 73 | // ----------------------------------------------------- |
| 74 | // '*' = match 0 or more occurances of anything |
| 75 | // "[abc]" = match anyof "abc" (ranges supported) |
| 76 | // "{xx,yy,zz}" = match anyof "xx", "yy", or "zz" |
| 77 | // '?' = match any character |
| 78 | // |
| 79 | // '\' is used to "escape" special characters |
| 80 | // Recursive |
| 81 | |
| 82 | bool |
| 83 | wxMatchWild (const char *pattern, const char *str, bool dot_special) |
| 84 | { |
| 85 | char c; |
| 86 | const char *cp; |
| 87 | bool done = FALSE, ret_code, ok; |
| 88 | // Below is for vi fans |
| 89 | const char OB = '{', CB = '}'; |
| 90 | |
| 91 | #if 0 |
| 92 | if (strcmp(pattern, "*.*") == 0) |
| 93 | pattern = "*"; // Hack for MS-DOS compat. |
| 94 | #endif |
| 95 | |
| 96 | // dot_special means '.' only matches '.' |
| 97 | if (dot_special && *str == '.' && *pattern != *str) |
| 98 | return FALSE; |
| 99 | |
| 100 | while ((*pattern != '\0') && (!done) && (((*str == '\0') && |
| 101 | ((*pattern == OB) || (*pattern == '*'))) || (*str != '\0'))) |
| 102 | { |
| 103 | switch (*pattern) |
| 104 | { |
| 105 | case '\\': |
| 106 | pattern++; |
| 107 | if (*pattern != '\0') |
| 108 | pattern++; |
| 109 | break; |
| 110 | case '*': |
| 111 | pattern++; |
| 112 | ret_code = FALSE; |
| 113 | while ((*str != '\0') && (!(ret_code = wxMatchWild (pattern, str++, FALSE)))); |
| 114 | if (ret_code) |
| 115 | { |
| 116 | while (*str != '\0') |
| 117 | str++; |
| 118 | while (*pattern != '\0') |
| 119 | pattern++; |
| 120 | } |
| 121 | break; |
| 122 | case '[': |
| 123 | pattern++; |
| 124 | repeat: |
| 125 | if ((*pattern == '\0') || (*pattern == ']')) |
| 126 | { |
| 127 | done = TRUE; |
| 128 | break; |
| 129 | } |
| 130 | if (*pattern == '\\') |
| 131 | { |
| 132 | pattern++; |
| 133 | if (*pattern == '\0') |
| 134 | { |
| 135 | done = TRUE; |
| 136 | break; |
| 137 | } |
| 138 | } |
| 139 | if (*(pattern + 1) == '-') |
| 140 | { |
| 141 | c = *pattern; |
| 142 | pattern += 2; |
| 143 | if (*pattern == ']') |
| 144 | { |
| 145 | done = TRUE; |
| 146 | break; |
| 147 | } |
| 148 | if (*pattern == '\\') |
| 149 | { |
| 150 | pattern++; |
| 151 | if (*pattern == '\0') |
| 152 | { |
| 153 | done = TRUE; |
| 154 | break; |
| 155 | } |
| 156 | } |
| 157 | if ((*str < c) || (*str > *pattern)) |
| 158 | { |
| 159 | pattern++; |
| 160 | goto repeat; |
| 161 | } |
| 162 | } |
| 163 | else if (*pattern != *str) |
| 164 | { |
| 165 | pattern++; |
| 166 | goto repeat; |
| 167 | } |
| 168 | pattern++; |
| 169 | while ((*pattern != ']') && (*pattern != '\0')) |
| 170 | { |
| 171 | if ((*pattern == '\\') && (*(pattern + 1) != '\0')) |
| 172 | pattern++; |
| 173 | pattern++; |
| 174 | } // while() |
| 175 | if (*pattern != '\0') |
| 176 | { |
| 177 | pattern++, str++; |
| 178 | } |
| 179 | break; |
| 180 | case '?': |
| 181 | pattern++; |
| 182 | str++; |
| 183 | break; |
| 184 | case OB: |
| 185 | pattern++; |
| 186 | while ((*pattern != CB) && (*pattern != '\0')) |
| 187 | { |
| 188 | cp = str; |
| 189 | ok = TRUE; |
| 190 | while (ok && (*cp != '\0') && (*pattern != '\0') && |
| 191 | (*pattern != ',') && (*pattern != CB)) |
| 192 | { |
| 193 | if (*pattern == '\\') |
| 194 | pattern++; |
| 195 | ok = (*pattern++ == *cp++); |
| 196 | } // while() |
| 197 | if (*pattern == '\0') |
| 198 | { |
| 199 | ok = FALSE; |
| 200 | done = TRUE; |
| 201 | break; |
| 202 | } |
| 203 | else if (ok) |
| 204 | { |
| 205 | str = cp; |
| 206 | while ((*pattern != CB) && (*pattern != '\0')) |
| 207 | { |
| 208 | if (*++pattern == '\\') |
| 209 | { |
| 210 | if (*++pattern == CB) |
| 211 | pattern++; |
| 212 | } |
| 213 | } // while() |
| 214 | } |
| 215 | else |
| 216 | { |
| 217 | while (*pattern != CB && *pattern != ',' && *pattern != '\0') |
| 218 | { |
| 219 | if (*++pattern == '\\') |
| 220 | { |
| 221 | if (*++pattern == CB || *pattern == ',') |
| 222 | pattern++; |
| 223 | } |
| 224 | } // while() |
| 225 | } |
| 226 | if (*pattern != '\0') |
| 227 | pattern++; |
| 228 | } // while() |
| 229 | break; |
| 230 | default: |
| 231 | if (*str == *pattern) |
| 232 | { |
| 233 | str++, pattern++; |
| 234 | } |
| 235 | else |
| 236 | { |
| 237 | done = TRUE; |
| 238 | } |
| 239 | } // switch() |
| 240 | } // while() |
| 241 | while (*pattern == '*') |
| 242 | pattern++; |
| 243 | return ((*str == '\0') && (*pattern == '\0')); |
| 244 | } |
| 245 | |
| 246 | #else /* MS-DOS/Windows glob() */ |
| 247 | /************************************************************************* |
| 248 | * |
| 249 | * wxMatchWild matches the given pattern string against |
| 250 | * a text string, and returns TRUE if it matches, FALSE otherwise. |
| 251 | * |
| 252 | * A match means that the entire text string is used up in the matching. |
| 253 | * The pattern can contain the following wildcards. |
| 254 | * |
| 255 | * * -- matches any sequence of characters |
| 256 | * ? -- matches one character |
| 257 | * |
| 258 | * If one or other or both of the string arguments to wxMatchWild function is |
| 259 | * NULL (i.e. there isn't a string), then the function returns FALSE. |
| 260 | * |
| 261 | */ |
| 262 | static bool wxPatternMatch (const char *pattern, const char *text, size_t i, size_t j); |
| 263 | |
| 264 | // @@@@ dotSpecial is ignored by MS-DOS |
| 265 | bool |
| 266 | wxMatchWild (const char *pattern, const char *text, bool /* dotSpecial */ ) |
| 267 | { |
| 268 | if (pattern == NULL || text == NULL || *pattern == '\0' || *text == '\0') |
| 269 | return FALSE; |
| 270 | return wxPatternMatch (pattern, text, 0, 0); |
| 271 | } |
| 272 | |
| 273 | /************************************************************************* |
| 274 | * |
| 275 | * wxPatternMatch does the work for wxMatchWild. wxPatternMatch matches |
| 276 | * the given pattern string against a text string, and returns TRUE if |
| 277 | * it matches, FALSE otherwise. It is assumed that the string arguments |
| 278 | * to wxPatternMatch exist. |
| 279 | * |
| 280 | * A match means that the entire text string is used up in the matching. |
| 281 | * The pattern can contain the following wildcards. |
| 282 | * |
| 283 | * * -- matches any sequence of characters |
| 284 | * ? -- matches one character |
| 285 | * |
| 286 | * wxPatternMatch works by going down the pattern trying to match the |
| 287 | * the same index character in the pattern and string arrays, and stops |
| 288 | * when the end of the pattern or text string is reached. However, if a |
| 289 | * '*' wildcard is met, the algorithm checks to see whether the remaining |
| 290 | * pattern (after the wildcard) matches the rest of the text (i.e. the |
| 291 | * wxPatternMatch function is called recursively). |
| 292 | */ |
| 293 | // Recursive |
| 294 | static bool |
| 295 | wxPatternMatch (const char *pattern, const char *text, size_t i, size_t j) |
| 296 | { |
| 297 | size_t pattern_length = strlen (pattern); |
| 298 | size_t text_length = strlen (text); |
| 299 | bool match = FALSE; |
| 300 | |
| 301 | #ifdef wx_msw |
| 302 | // MS-DOS file system is case INDEPENDENT |
| 303 | # define EQU(x,y) (wxToLower(x) == wxToLower(y)) |
| 304 | #else |
| 305 | # define EQU(x,y) ((x) == (y)) |
| 306 | #endif |
| 307 | |
| 308 | while (j < pattern_length && i < text_length) |
| 309 | { |
| 310 | if (EQU(text[i], pattern[j]) || pattern[j] == '?') |
| 311 | { |
| 312 | match = TRUE; |
| 313 | i++, j++; |
| 314 | } |
| 315 | else if (pattern[j] == '*') |
| 316 | { |
| 317 | // If pattern ends in '*' |
| 318 | if (++j == pattern_length) |
| 319 | { |
| 320 | match = TRUE; |
| 321 | i = text_length; |
| 322 | } |
| 323 | else |
| 324 | { |
| 325 | match = FALSE; |
| 326 | // after wildcard check to see whether rest of pattern matches |
| 327 | // up with rest of text |
| 328 | while (i < text_length && match != TRUE) |
| 329 | { |
| 330 | match = wxPatternMatch (pattern, text, i, j); |
| 331 | i++; |
| 332 | } |
| 333 | // text index is decremented so that it points to where |
| 334 | // the text string starts to match the rest of the pattern |
| 335 | i--; |
| 336 | } |
| 337 | } |
| 338 | else if (! EQU(text[i], pattern[j])) |
| 339 | { |
| 340 | j = pattern_length; |
| 341 | match = FALSE; |
| 342 | } |
| 343 | } |
| 344 | if (j == pattern_length && i == text_length && match == TRUE) |
| 345 | { |
| 346 | return TRUE; |
| 347 | } |
| 348 | else |
| 349 | // special case where pattern and text are the same except that pattern |
| 350 | // also only has '*' wildcards on the end |
| 351 | if (i == text_length && pattern[j] == '*' && match == TRUE) |
| 352 | { |
| 353 | for (; j < pattern_length; j++) |
| 354 | { |
| 355 | if (pattern[j] != '*') |
| 356 | return FALSE; |
| 357 | } |
| 358 | return TRUE; |
| 359 | } |
| 360 | else |
| 361 | { |
| 362 | return FALSE; |
| 363 | } |
| 364 | } |
| 365 | |
| 366 | #endif /* UNIX_GLOB */ |
| 367 | //----------------------------------------------------------------------------- |