wimlib.net Git - wimlib/blob - src/wildcard.c

   1 /*
   2  * wildcard.c
   3  *
   4  * Wildcard matching functions.
   5  */
   6
   7 /*
   8  * Copyright (C) 2013 Eric Biggers
   9  *
  10  * This file is free software; you can redistribute it and/or modify it under
  11  * the terms of the GNU Lesser General Public License as published by the Free
  12  * Software Foundation; either version 3 of the License, or (at your option) any
  13  * later version.
  14  *
  15  * This file is distributed in the hope that it will be useful, but WITHOUT
  16  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  17  * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  18  * details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public License
  21  * along with this file; if not, see http://www.gnu.org/licenses/.
  22  */
  23
  24 #ifdef HAVE_CONFIG_H
  25 #  include "config.h"
  26 #endif
  27
  28 #include <ctype.h>
  29
  30 #include "wimlib/dentry.h"
  31 #include "wimlib/encoding.h"
  32 #include "wimlib/error.h"
  33 #include "wimlib/metadata.h"
  34 #include "wimlib/paths.h"
  35 #include "wimlib/wildcard.h"
  36
  37 struct match_dentry_ctx {
  38         int (*consume_dentry)(struct wim_dentry *, void *);
  39         void *consume_dentry_ctx;
  40         size_t consume_dentry_count;
  41         tchar *wildcard_path;
  42         size_t cur_component_offset;
  43         size_t cur_component_len;
  44         bool case_insensitive;
  45 };
  46
  47 static bool
  48 do_match_wildcard(const tchar *string, size_t string_len,
  49                   const tchar *wildcard, size_t wildcard_len,
  50                   bool ignore_case)
  51 {
  52         for (;;) {
  53                 if (string_len == 0) {
  54                         while (wildcard_len != 0 && *wildcard == T('*')) {
  55                                 wildcard++;
  56                                 wildcard_len--;
  57                         }
  58                         return (wildcard_len == 0);
  59                 } else if (wildcard_len == 0) {
  60                         return false;
  61                 } else if (*string == *wildcard || *wildcard == T('?') ||
  62                            (ignore_case && totlower(*string) == totlower(*wildcard)))
  63                 {
  64                         string++;
  65                         string_len--;
  66                         wildcard_len--;
  67                         wildcard++;
  68                         continue;
  69                 } else if (*wildcard == T('*')) {
  70                         return do_match_wildcard(string, string_len,
  71                                                  wildcard + 1, wildcard_len - 1,
  72                                                  ignore_case) ||
  73                                do_match_wildcard(string + 1, string_len - 1,
  74                                                  wildcard, wildcard_len,
  75                                                  ignore_case);
  76                 } else {
  77                         return false;
  78                 }
  79         }
  80 }
  81
  82 static bool
  83 match_wildcard(const tchar *string, const tchar *wildcard,
  84                size_t wildcard_len, bool ignore_case)
  85 {
  86         return do_match_wildcard(string, tstrlen(string),
  87                                  wildcard, wildcard_len, ignore_case);
  88 }
  89
  90 /*
  91  * Determines whether a path matches a wildcard pattern.
  92  *
  93  * @path
  94  *      The path to match.  Assumptions:  All path separators must be @path_sep,
  95  *      there cannot be consecutive path separators, there cannot be a trailing
  96  *      path separator, and there must be exactly one leading path separator.
  97  *
  98  * @path_nchars
  99  *      Number of characters in @path.
 100  *
 101  * @wildcard
 102  *      The wildcard pattern to match.  It can contain the wildcard characters
 103  *      '*' and '?'.  The former matches zero or more characters except
 104  *      @path_sep, and the latter matches any character except @path_sep.  All
 105  *      path separators in the pattern must be @path_sep, and there cannot be
 106  *      consecutive path separators, and there cannot be a trailing path
 107  *      separator.  If there is a leading path separator, the match is attempted
 108  *      with the filename only; otherwise, the match is attempted with the whole
 109  *      path.
 110  *
 111  * @path_sep
 112  *      Path separator character used in @path and @wildcard.
 113  *
 114  * @prefix_ok
 115  *      If %true, allow a prefix of @path, terminated by a path separator, to
 116  *      match the pattern, in addition to @path itself.  In other words, return
 117  *      %true if the pattern actually matches one of the ancestor directories of
 118  *      @path.
 119  *
 120  * Returns %true if there was a match; %false if there was not.
 121  */
 122 bool
 123 match_path(const tchar *path, size_t path_nchars,
 124            const tchar *wildcard, tchar path_sep, bool prefix_ok)
 125 {
 126         if (*wildcard != path_sep) {
 127                 /* Pattern doesn't begin with path separator.  Try to match the
 128                  * file name only.  */
 129                 return match_wildcard(path_basename_with_len(path, path_nchars),
 130                                       wildcard, tstrlen(wildcard),
 131                                       default_ignore_case);
 132         } else {
 133                 /* Pattern begins with path separator.  Try to match the whole
 134                  * path.  */
 135                 do {
 136                         if (!*wildcard) {
 137                                 /* Path has more components than pattern  */
 138                                 return prefix_ok;
 139                         }
 140
 141                         size_t path_component_len = 0;
 142                         size_t wildcard_component_len = 0;
 143
 144                         do {
 145                                 path_component_len++;
 146                         } while (path[path_component_len] != path_sep &&
 147                                  path[path_component_len] != T('\0'));
 148                         do {
 149                                 wildcard_component_len++;
 150                         } while (wildcard[wildcard_component_len] != path_sep &&
 151                                  wildcard[wildcard_component_len] != T('\0'));
 152                         if (!do_match_wildcard(path, path_component_len,
 153                                                wildcard, wildcard_component_len,
 154                                                default_ignore_case))
 155                                 return false;
 156                         path += path_component_len;
 157                         wildcard += wildcard_component_len;
 158                 } while (*path);
 159
 160                 return (*wildcard == '\0');
 161         }
 162 }
 163
 164 static int
 165 expand_wildcard_recursive(struct wim_dentry *cur_dentry,
 166                           struct match_dentry_ctx *ctx);
 167
 168 enum {
 169         WILDCARD_STATUS_DONE_FULLY,
 170         WILDCARD_STATUS_DONE_TRAILING_SLASHES,
 171         WILDCARD_STATUS_NOT_DONE,
 172 };
 173
 174 static int
 175 wildcard_status(const tchar *wildcard)
 176 {
 177         if (*wildcard == T('\0'))
 178                 return WILDCARD_STATUS_DONE_FULLY;
 179         while (*wildcard == WIM_PATH_SEPARATOR)
 180                 wildcard++;
 181         if (*wildcard == T('\0'))
 182                 return WILDCARD_STATUS_DONE_TRAILING_SLASHES;
 183
 184         return WILDCARD_STATUS_NOT_DONE;
 185 }
 186
 187 static int
 188 match_dentry(struct wim_dentry *cur_dentry, struct match_dentry_ctx *ctx)
 189 {
 190         const tchar *name;
 191         size_t name_nchars;
 192         int ret;
 193
 194         if (cur_dentry->d_name_nbytes == 0)
 195                 return 0;
 196
 197         ret = utf16le_get_tstr(cur_dentry->d_name,
 198                                cur_dentry->d_name_nbytes,
 199                                &name, &name_nchars);
 200         if (ret)
 201                 return ret;
 202         name_nchars /= sizeof(tchar);
 203
 204         if (do_match_wildcard(name,
 205                               name_nchars,
 206                               &ctx->wildcard_path[ctx->cur_component_offset],
 207                               ctx->cur_component_len,
 208                               ctx->case_insensitive))
 209         {
 210                 switch (wildcard_status(&ctx->wildcard_path[
 211                                 ctx->cur_component_offset +
 212                                 ctx->cur_component_len]))
 213                 {
 214                 case WILDCARD_STATUS_DONE_TRAILING_SLASHES:
 215                         if (!dentry_is_directory(cur_dentry)) {
 216                                 ret = 0;
 217                                 break;
 218                         }
 219                         /* Fall through  */
 220                 case WILDCARD_STATUS_DONE_FULLY:
 221                         ret = (*ctx->consume_dentry)(cur_dentry,
 222                                                      ctx->consume_dentry_ctx);
 223                         ctx->consume_dentry_count++;
 224                         break;
 225                 case WILDCARD_STATUS_NOT_DONE:
 226                         ret = expand_wildcard_recursive(cur_dentry, ctx);
 227                         break;
 228                 }
 229         } else {
 230                 ret = 0;
 231         }
 232
 233         utf16le_put_tstr(name);
 234
 235         return ret;
 236 }
 237
 238 static int
 239 expand_wildcard_recursive(struct wim_dentry *cur_dentry,
 240                           struct match_dentry_ctx *ctx)
 241 {
 242         tchar *w;
 243         size_t begin;
 244         size_t end;
 245         size_t len;
 246         size_t offset_save;
 247         size_t len_save;
 248         int ret;
 249         struct wim_dentry *child;
 250
 251         w = ctx->wildcard_path;
 252
 253         begin = ctx->cur_component_offset + ctx->cur_component_len;
 254         while (w[begin] == WIM_PATH_SEPARATOR)
 255                 begin++;
 256
 257         end = begin;
 258
 259         while (w[end] != T('\0') && w[end] != WIM_PATH_SEPARATOR)
 260                 end++;
 261
 262         len = end - begin;
 263
 264         if (len == 0)
 265                 return 0;
 266
 267         offset_save = ctx->cur_component_offset;
 268         len_save = ctx->cur_component_len;
 269
 270         ctx->cur_component_offset = begin;
 271         ctx->cur_component_len = len;
 272
 273         ret = 0;
 274         for_dentry_child(child, cur_dentry) {
 275                 ret = match_dentry(child, ctx);
 276                 if (ret)
 277                         break;
 278         }
 279
 280         ctx->cur_component_len = len_save;
 281         ctx->cur_component_offset = offset_save;
 282
 283         return ret;
 284 }
 285
 286 /* Expand a wildcard relative to the current WIM image.
 287  *
 288  * @wim
 289  *      WIMStruct whose currently selected image is searched to expand the
 290  *      wildcard.
 291  * @wildcard_path
 292  *      Wildcard path to expand, which may contain the '?' and '*' characters.
 293  *      Path separators must be WIM_PATH_SEPARATOR.  Leading path separators are
 294  *      ignored, whereas one or more trailing path separators indicate that the
 295  *      wildcard path can only match directories (and not reparse points).
 296  * @consume_dentry
 297  *      Callback function which will receive each directory entry matched by the
 298  *      wildcard.
 299  * @consume_dentry_ctx
 300  *      Argument to pass to @consume_dentry.
 301  * @flags
 302  *      Zero or more of the following flags:
 303  *
 304  *      WILDCARD_FLAG_WARN_IF_NO_MATCH:
 305  *              Issue a warning if the wildcard does not match any dentries.
 306  *
 307  *      WILDCARD_FLAG_ERROR_IF_NO_MATCH:
 308  *              Issue an error and return WIMLIB_ERR_PATH_DOES_NOT_EXIST if the
 309  *              wildcard does not match any dentries.
 310  *
 311  *      WILDCARD_FLAG_CASE_INSENSITIVE:
 312  *              Perform the matching case insensitively.  Note that this may
 313  *              cause @wildcard to match multiple dentries, even if it does not
 314  *              contain wildcard characters.
 315  *
 316  * @return 0 on success; a positive error code on error; or the first nonzero
 317  * value returned by @consume_dentry.
 318  */
 319 int
 320 expand_wildcard(WIMStruct *wim,
 321                 const tchar *wildcard_path,
 322                 int (*consume_dentry)(struct wim_dentry *, void *),
 323                 void *consume_dentry_ctx,
 324                 u32 flags)
 325 {
 326         struct wim_dentry *root;
 327         int ret;
 328
 329         root = wim_get_current_root_dentry(wim);
 330         if (root == NULL)
 331                 goto no_match;
 332
 333         struct match_dentry_ctx ctx = {
 334                 .consume_dentry = consume_dentry,
 335                 .consume_dentry_ctx = consume_dentry_ctx,
 336                 .consume_dentry_count = 0,
 337                 .wildcard_path = TSTRDUP(wildcard_path),
 338                 .cur_component_offset = 0,
 339                 .cur_component_len = 0,
 340                 .case_insensitive = ((flags & WILDCARD_FLAG_CASE_INSENSITIVE) != 0),
 341         };
 342
 343         if (ctx.wildcard_path == NULL)
 344                 return WIMLIB_ERR_NOMEM;
 345
 346         ret = expand_wildcard_recursive(root, &ctx);
 347         FREE(ctx.wildcard_path);
 348         if (ret == 0 && ctx.consume_dentry_count == 0)
 349                 goto no_match;
 350         return ret;
 351
 352 no_match:
 353         ret = 0;
 354         if (flags & WILDCARD_FLAG_WARN_IF_NO_MATCH)
 355                 WARNING("No matches for wildcard path \"%"TS"\"", wildcard_path);
 356
 357         if (flags & WILDCARD_FLAG_ERROR_IF_NO_MATCH) {
 358                 ERROR("No matches for wildcard path \"%"TS"\"", wildcard_path);
 359                 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
 360         }
 361         return ret;
 362 }