2 * capture_common.c - Mostly code to handle excluding paths from capture.
6 * Copyright (C) 2013, 2014 Eric Biggers
8 * This file is part of wimlib, a library for working with WIM files.
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)
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
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/.
28 #include "wimlib/capture.h"
29 #include "wimlib/dentry.h"
30 #include "wimlib/error.h"
31 #include "wimlib/lookup_table.h"
32 #include "wimlib/paths.h"
33 #include "wimlib/progress.h"
34 #include "wimlib/textfile.h"
35 #include "wimlib/wildcard.h"
40 * Tally a file (or directory) that has been scanned for a capture operation,
41 * and possibly call the progress function provided by the library user.
44 * Flags, optional progress function, and progress data for the capture
47 * Status of the scanned file (ok, unsupported, excluded, or excluded
50 * If @status is WIMLIB_SCAN_DENTRY_OK, this is a pointer to the WIM inode
51 * that has been created for the scanned file. The first time the file is
52 * seen, inode->i_nlink will be 1. On subsequent visits of the same inode
53 * via additional hard links, inode->i_nlink will be greater than 1.
56 do_capture_progress(struct add_image_params *params, int status,
57 const struct wim_inode *inode)
60 case WIMLIB_SCAN_DENTRY_OK:
61 if (!(params->add_flags & WIMLIB_ADD_FLAG_VERBOSE))
63 case WIMLIB_SCAN_DENTRY_UNSUPPORTED:
64 case WIMLIB_SCAN_DENTRY_EXCLUDED:
65 case WIMLIB_SCAN_DENTRY_EXCLUDED_SYMLINK:
66 case WIMLIB_SCAN_DENTRY_FIXED_SYMLINK:
67 if (!(params->add_flags & WIMLIB_ADD_FLAG_EXCLUDE_VERBOSE))
70 params->progress.scan.status = status;
71 if (status == WIMLIB_SCAN_DENTRY_OK && inode->i_nlink == 1) {
73 /* Successful scan, and visiting inode for the first time */
75 /* Tally size of all data streams. */
76 const struct wim_lookup_table_entry *lte;
77 for (unsigned i = 0; i <= inode->i_num_ads; i++) {
78 lte = inode_stream_lte_resolved(inode, i);
80 params->progress.scan.num_bytes_scanned += lte->size;
83 /* Tally the file itself. */
84 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
85 params->progress.scan.num_dirs_scanned++;
87 params->progress.scan.num_nondirs_scanned++;
90 /* Call the user-provided progress function. */
91 return call_progress(params->progfunc, WIMLIB_PROGRESS_MSG_SCAN_DENTRY,
92 ¶ms->progress, params->progctx);
96 * Given a null-terminated pathname pattern @pat that has been read from line
97 * @line_no of the file @path, validate and canonicalize the pattern.
99 * On success, returns 0.
100 * On failure, returns WIMLIB_ERR_INVALID_CAPTURE_CONFIG.
101 * In either case, @pat may have been modified in-place (and possibly
105 mangle_pat(tchar *pat, const tchar *path, unsigned long line_no)
107 if (!is_any_path_separator(pat[0]) &&
108 pat[0] != T('\0') && pat[1] == T(':'))
110 /* Pattern begins with drive letter. */
112 if (!is_any_path_separator(pat[2])) {
113 /* Something like c:file, which is actually a path
114 * relative to the current working directory on the c:
115 * drive. We require paths with drive letters to be
117 ERROR("%"TS":%lu: Invalid pattern \"%"TS"\":\n"
118 " Patterns including drive letters must be absolute!\n"
119 " Maybe try \"%"TC":%"TC"%"TS"\"?\n",
121 pat[0], OS_PREFERRED_PATH_SEPARATOR, &pat[2]);
122 return WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
125 WARNING("%"TS":%lu: Pattern \"%"TS"\" starts with a drive "
126 "letter, which is being removed.",
129 /* Strip the drive letter. */
130 tmemmove(pat, pat + 2, tstrlen(pat + 2) + 1);
133 /* Collapse consecutive path separators, and translate both / and \ into
134 * / (UNIX) or \ (Windows).
136 * Note: we expect that this function produces patterns that can be used
137 * for both filesystem paths and WIM paths, so the desired path
138 * separators must be the same. */
139 BUILD_BUG_ON(OS_PREFERRED_PATH_SEPARATOR != WIM_PATH_SEPARATOR);
140 do_canonicalize_path(pat, pat);
142 /* Relative patterns can only match file names, so they must be
143 * single-component only. */
144 if (pat[0] != OS_PREFERRED_PATH_SEPARATOR &&
145 tstrchr(pat, OS_PREFERRED_PATH_SEPARATOR))
147 ERROR("%"TS":%lu: Invalid pattern \"%"TS"\":\n"
148 " Relative patterns can only include one path component!\n"
149 " Maybe try \"%"TC"%"TS"\"?",
150 path, line_no, pat, OS_PREFERRED_PATH_SEPARATOR, pat);
151 return WIMLIB_ERR_INVALID_CAPTURE_CONFIG;
158 * Read, parse, and validate a capture configuration file from either an on-disk
159 * file or an in-memory buffer.
161 * To read from a file, specify @config_file, and use NULL for @buf.
162 * To read from a buffer, specify @buf and @bufsize.
164 * @config must be initialized to all 0's.
166 * On success, 0 will be returned, and the resulting capture configuration will
167 * be stored in @config.
169 * On failure, a positive error code will be returned, and the contents of
170 * @config will be invalidated.
173 read_capture_config(const tchar *config_file, const void *buf,
174 size_t bufsize, struct capture_config *config)
178 /* [PrepopulateList] is used for apply, not capture. But since we do
179 * understand it, recognize it, thereby avoiding the unrecognized
180 * section warning, but discard the resulting strings. */
181 STRING_SET(prepopulate_pats);
183 struct text_file_section sections[] = {
185 &config->exclusion_pats},
186 {T("ExclusionException"),
187 &config->exclusion_exception_pats},
188 {T("PrepopulateList"),
193 ret = do_load_text_file(config_file, buf, bufsize, &mem,
194 sections, ARRAY_LEN(sections),
195 LOAD_TEXT_FILE_REMOVE_QUOTES, mangle_pat);
199 FREE(prepopulate_pats.strings);
206 destroy_capture_config(struct capture_config *config)
208 FREE(config->exclusion_pats.strings);
209 FREE(config->exclusion_exception_pats.strings);
214 * Determine whether a path matches any wildcard pattern in a list.
216 * Special rules apply about what form @path must be in; see match_path().
219 match_pattern_list(const tchar *path, size_t path_nchars,
220 const struct string_set *list)
222 for (size_t i = 0; i < list->num_strings; i++)
223 if (match_path(path, path_nchars, list->strings[i],
224 OS_PREFERRED_PATH_SEPARATOR, true))
230 * Determine whether the filesystem @path should be excluded from capture, based
231 * on the current capture configuration file.
233 * The @path must be given relative to the root of the capture, but with a
234 * leading path separator. For example, if the file "in/file" is being tested
235 * and the library user ran wimlib_add_image(wim, "in", ...), then the directory
236 * "in" is the root of the capture and the path should be specified as "/file".
238 * Also, all path separators in @path must be OS_PREFERRED_PATH_SEPARATOR, there
239 * cannot be trailing slashes, and there cannot be consecutive path separators.
241 * As a special case, the empty string will be interpreted as a single path
242 * separator (which means the root of capture itself).
245 should_exclude_path(const tchar *path, size_t path_nchars,
246 const struct capture_config *config)
254 dummy[0] = OS_PREFERRED_PATH_SEPARATOR;
260 return match_pattern_list(path, path_nchars, &config->exclusion_pats) &&
261 !match_pattern_list(path, path_nchars, &config->exclusion_exception_pats);