X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fcapture_common.c;h=8279497113280b9a82e53882ceee0d08ce9a802a;hp=0fa71adcbd898c5980a85ff42138cc3bef06f3ae;hb=edcb6f0d4a843d4e05a7b7e3b72006ac4b4acf6f;hpb=c86d5f74e13e5aeff57aff84f7d65264607a29ac diff --git a/src/capture_common.c b/src/capture_common.c index 0fa71adc..82794971 100644 --- a/src/capture_common.c +++ b/src/capture_common.c @@ -67,20 +67,26 @@ do_capture_progress(struct capture_params *params, int status, break; } params->progress.scan.status = status; - if (status == WIMLIB_SCAN_DENTRY_OK && inode->i_nlink == 1) { - - /* Successful scan, and visiting inode for the first time */ - - /* Tally size of all streams. */ - for (unsigned i = 0; i < inode->i_num_streams; i++) { - const struct blob_descriptor *blob = - stream_blob_resolved(&inode->i_streams[i]); - if (blob) - params->progress.scan.num_bytes_scanned += blob->size; + if (status == WIMLIB_SCAN_DENTRY_OK) { + + /* The first time the inode is seen, tally all its streams. */ + if (inode->i_nlink == 1) { + for (unsigned i = 0; i < inode->i_num_streams; i++) { + const struct blob_descriptor *blob = + stream_blob_resolved(&inode->i_streams[i]); + if (blob) + params->progress.scan.num_bytes_scanned += blob->size; + } } - /* Tally the file itself. */ - if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) + /* Tally the file itself, counting every hard link. It's + * debatable whether every link should be counted, but counting + * every link makes the statistics consistent with the ones + * placed in the FILECOUNT and DIRCOUNT elements of the WIM + * file's XML document. It also avoids possible user confusion + * if the number of files reported were to be lower than that + * displayed by some other software such as file browsers. */ + if (inode_is_directory(inode)) params->progress.scan.num_dirs_scanned++; else params->progress.scan.num_nondirs_scanned++; @@ -135,7 +141,7 @@ mangle_pat(tchar *pat, const tchar *path, unsigned long line_no) * Note: we expect that this function produces patterns that can be used * for both filesystem paths and WIM paths, so the desired path * separators must be the same. */ - BUILD_BUG_ON(OS_PREFERRED_PATH_SEPARATOR != WIM_PATH_SEPARATOR); + STATIC_ASSERT(OS_PREFERRED_PATH_SEPARATOR == WIM_PATH_SEPARATOR); do_canonicalize_path(pat, pat); /* Relative patterns can only match file names, so they must be @@ -307,3 +313,51 @@ try_exclude(const tchar *full_path, const struct capture_params *params) return 0; } + +/* + * Determine whether a directory entry of the specified name should be ignored. + * This is a lower level function which runs prior to try_exclude(). It handles + * the standard '.' and '..' entries, which show up in directory listings but + * should not be archived. It also checks for odd filenames that usually should + * not exist but could cause problems if archiving them were to be attempted. + */ +bool +should_ignore_filename(const tchar *name, const int name_nchars) +{ + if (name_nchars <= 0) { + WARNING("Ignoring empty filename"); + return true; + } + + if (name[0] == T('.') && + (name_nchars == 1 || (name_nchars == 2 && name[1] == T('.')))) + return true; + + for (int i = 0; i < name_nchars; i++) { + if (name[i] == T('\0')) { + WARNING("Ignoring filename containing embedded null character"); + return true; + } + if (name[i] == OS_PREFERRED_PATH_SEPARATOR) { + WARNING("Ignoring filename containing embedded path separator"); + return true; + } + } + + return false; +} + +/* Attach a newly scanned directory tree to its parent directory, with duplicate + * handling. */ +void +attach_scanned_tree(struct wim_dentry *parent, struct wim_dentry *child, + struct blob_table *blob_table) +{ + struct wim_dentry *duplicate; + + if (child && (duplicate = dentry_add_child(parent, child))) { + WARNING("Duplicate file path: \"%"TS"\". Only capturing " + "the first version.", dentry_full_path(duplicate)); + free_dentry_tree(child, blob_table); + } +}