Refactor headers
[wimlib] / src / capture_common.c
1 /*
2  * capture_common.c - Mostly code to handle excluding paths from capture.
3  */
4
5 /*
6  * Copyright (C) 2013 Eric Biggers
7  *
8  * This file is part of wimlib, a library for working with WIM files.
9  *
10  * wimlib is free software; you can redistribute it and/or modify it under the
11  * terms of the GNU General Public License as published by the Free
12  * Software Foundation; either version 3 of the License, or (at your option)
13  * any later version.
14  *
15  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
16  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
17  * A PARTICULAR PURPOSE. See the GNU General Public License for more
18  * details.
19  *
20  * You should have received a copy of the GNU General Public License
21  * along with wimlib; if not, see http://www.gnu.org/licenses/.
22  */
23
24 #ifdef HAVE_CONFIG_H
25 #  include "config.h"
26 #endif
27
28 #include "wimlib/assert.h"
29 #include "wimlib/capture.h"
30 #include "wimlib/error.h"
31 #include "wimlib/paths.h"
32
33 #ifdef __WIN32__
34 #  include "wimlib/win32.h" /* for fnmatch() equivalent */
35 #else
36 #  include <fnmatch.h>
37 #endif
38 #include <string.h>
39
40
41 static int
42 canonicalize_pattern(const tchar *pat, tchar **canonical_pat_ret)
43 {
44         tchar *canonical_pat;
45
46         if (pat[0] != T('/') && pat[0] != T('\\') &&
47             pat[0] != T('\0') && pat[1] == T(':'))
48         {
49                 /* Pattern begins with drive letter */
50                 if (pat[2] != T('/') && pat[2] != T('\\')) {
51                         /* Something like c:file, which is actually a path
52                          * relative to the current working directory on the c:
53                          * drive.  We require paths with drive letters to be
54                          * absolute. */
55                         ERROR("Invalid path \"%"TS"\"; paths including drive letters "
56                               "must be absolute!", pat);
57                         ERROR("Maybe try \"%"TC":/%"TS"\"?",
58                               pat[0], pat + 2);
59                         return WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
60                 }
61
62                 WARNING("Pattern \"%"TS"\" starts with a drive letter, which is "
63                         "being removed.", pat);
64                 /* Strip the drive letter */
65                 pat += 2;
66         }
67         canonical_pat = canonicalize_fs_path(pat);
68         if (!canonical_pat)
69                 return WIMLIB_ERR_NOMEM;
70         *canonical_pat_ret = canonical_pat;
71         return 0;
72 }
73
74 static int
75 copy_and_canonicalize_pattern_list(const struct wimlib_pattern_list *list,
76                                    struct wimlib_pattern_list *copy)
77 {
78         int ret = 0;
79
80         copy->pats = CALLOC(list->num_pats, sizeof(list->pats[0]));
81         if (!copy->pats)
82                 return WIMLIB_ERR_NOMEM;
83         copy->num_pats = list->num_pats;
84         for (size_t i = 0; i < list->num_pats; i++) {
85                 ret = canonicalize_pattern(list->pats[i], &copy->pats[i]);
86                 if (ret)
87                         break;
88         }
89         return ret;
90 }
91
92 int
93 copy_and_canonicalize_capture_config(const struct wimlib_capture_config *config,
94                                      struct wimlib_capture_config **config_copy_ret)
95 {
96         struct wimlib_capture_config *config_copy;
97         int ret;
98
99         config_copy = CALLOC(1, sizeof(struct wimlib_capture_config));
100         if (!config_copy) {
101                 ret = WIMLIB_ERR_NOMEM;
102                 goto out_free_capture_config;
103         }
104         ret = copy_and_canonicalize_pattern_list(&config->exclusion_pats,
105                                                  &config_copy->exclusion_pats);
106         if (ret)
107                 goto out_free_capture_config;
108         ret = copy_and_canonicalize_pattern_list(&config->exclusion_exception_pats,
109                                                  &config_copy->exclusion_exception_pats);
110         if (ret)
111                 goto out_free_capture_config;
112         *config_copy_ret = config_copy;
113         goto out;
114 out_free_capture_config:
115         free_capture_config(config_copy);
116 out:
117         return ret;
118 }
119
120 static void
121 destroy_pattern_list(struct wimlib_pattern_list *list)
122 {
123         for (size_t i = 0; i < list->num_pats; i++)
124                 FREE(list->pats[i]);
125         FREE(list->pats);
126 }
127
128 void
129 free_capture_config(struct wimlib_capture_config *config)
130 {
131         if (config) {
132                 destroy_pattern_list(&config->exclusion_pats);
133                 destroy_pattern_list(&config->exclusion_exception_pats);
134                 FREE(config);
135         }
136 }
137
138 static bool
139 match_pattern(const tchar *path,
140               const tchar *path_basename,
141               const struct wimlib_pattern_list *list)
142 {
143         for (size_t i = 0; i < list->num_pats; i++) {
144
145                 const tchar *pat = list->pats[i];
146                 const tchar *string;
147
148                 if (*pat == T('/')) {
149                         /* Absolute path from root of capture */
150                         string = path;
151                 } else {
152                         if (tstrchr(pat, T('/')))
153                                 /* Relative path from root of capture */
154                                 string = path + 1;
155                         else
156                                 /* A file name pattern */
157                                 string = path_basename;
158                 }
159
160                 /* Warning: on Windows native builds, fnmatch() calls the
161                  * replacement function in win32.c. */
162                 if (fnmatch(pat, string, FNM_PATHNAME | FNM_NOESCAPE
163                                 #ifdef FNM_CASEFOLD
164                                         | FNM_CASEFOLD
165                                 #endif
166                             ) == 0)
167                 {
168                         DEBUG("\"%"TS"\" matches the pattern \"%"TS"\"",
169                               string, pat);
170                         return true;
171                 } else {
172                         DEBUG2("\"%"TS"\" does not match the pattern \"%"TS"\"",
173                                string, pat);
174                 }
175         }
176         return false;
177 }
178
179 /* Return true if the image capture configuration file indicates we should
180  * exclude the filename @path from capture.
181  *
182  * If @exclude_prefix is %true, the part of the path up and including the name
183  * of the directory being captured is not included in the path for matching
184  * purposes.  This allows, for example, a pattern like /hiberfil.sys to match a
185  * file /mnt/windows7/hiberfil.sys if we are capturing the /mnt/windows7
186  * directory.
187  */
188 bool
189 exclude_path(const tchar *path, size_t path_len,
190              const struct wimlib_capture_config *config, bool exclude_prefix)
191 {
192         if (!config)
193                 return false;
194         const tchar *basename = path_basename_with_len(path, path_len);
195         if (exclude_prefix) {
196                 wimlib_assert(path_len >= config->_prefix_num_tchars);
197                 if (!tmemcmp(config->_prefix, path, config->_prefix_num_tchars) &&
198                     path[config->_prefix_num_tchars] == T('/'))
199                 {
200                         path += config->_prefix_num_tchars;
201                 }
202         }
203         return match_pattern(path, basename, &config->exclusion_pats) &&
204                 !match_pattern(path, basename, &config->exclusion_exception_pats);
205
206 }