RmTreeMerger *rm_tm_new(RmSession *session) {
RmTreeMerger *self = g_slice_new(RmTreeMerger);
self->session = session;
g_queue_init(&self->valid_dirs);
self->result_table = g_hash_table_new_full((GHashFunc)rm_directory_hash,
(GEqualFunc)rm_directory_equal, NULL,
(GDestroyNotify)g_queue_free);
self->file_groups =
g_hash_table_new_full((GHashFunc)rm_digest_hash, (GEqualFunc)rm_digest_equal,
NULL, (GDestroyNotify)g_queue_free);
self->file_checks = g_hash_table_new_full((GHashFunc)rm_digest_hash,
(GEqualFunc)rm_digest_equal, NULL, NULL);
self->known_hashs = g_hash_table_new_full(NULL, NULL, NULL, NULL);
rm_trie_init(&self->dir_tree);
rm_trie_init(&self->count_tree);
rm_tm_chunk_paths(self, session->cfg->paths);
return self;
}
int main(void) {
RmTrie trie;
rm_trie_init(&trie);
GTimer *timer = g_timer_new();
g_timer_start(timer);
char buf[1024];
int i = 0;
while(fgets(buf, sizeof(buf), stdin)) {
buf[strlen(buf) - 1] = 0;
rm_trie_insert(&trie, buf, GUINT_TO_POINTER(++i));
memset(buf, 0, sizeof(buf));
}
g_printerr("Took %2.5f to insert %d items\n", g_timer_elapsed(timer, NULL), i);
rm_trie_print(&trie);
memset(buf, 0, sizeof(buf));
rm_trie_build_path(&trie, rm_trie_search_node(&trie, "/usr/bin/rmlint"), buf,
sizeof(buf));
g_printerr("=> %s\n", buf);
g_timer_start(timer);
const int N = 10000000;
for(int x = 0; x < N; x++) {
rm_trie_search(&trie, "/usr/bin/rmlint");
}
g_printerr("Took %2.5f to search\n", g_timer_elapsed(timer, NULL));
g_printerr("%u\n", GPOINTER_TO_UINT(rm_trie_search(&trie, "/usr/bin/rmlint")));
g_printerr("%u\n", GPOINTER_TO_UINT(rm_trie_search(&trie, "/a/b/c")));
rm_trie_destroy(&trie);
g_timer_destroy(timer);
return 0;
}
/* Options not specified by commandline get a default option -
* this is usually called before rm_cmd_parse_args */
void rm_cfg_set_default(RmCfg *cfg) {
/* Set everything to 0 at first,
* only non-null options are listed below.
*/
memset(cfg, 0, sizeof(RmCfg));
/* Traversal options */
cfg->depth = PATH_MAX / 2;
cfg->minsize = 0;
cfg->maxsize = G_MAXUINT64;
/* Lint Types */
cfg->ignore_hidden = true;
cfg->find_emptydirs = true;
cfg->find_emptyfiles = true;
cfg->find_duplicates = true;
cfg->find_badids = true;
cfg->find_badlinks = true;
cfg->find_hardlinked_dupes = true;
cfg->build_fiemap = true;
cfg->crossdev = true;
cfg->list_mounts = true;
/* Misc options */
cfg->sort_criteria = g_strdup("pOma");
cfg->checksum_type = RM_DEFAULT_DIGEST;
cfg->with_color = true;
cfg->with_stdout_color = true;
cfg->with_stderr_color = true;
cfg->threads = 16;
cfg->threads_per_disk = 2;
cfg->verbosity = G_LOG_LEVEL_INFO;
cfg->follow_symlinks = false;
cfg->total_mem = (RmOff) 1024 * 1024 * 1024;
cfg->sweep_size = 1024 * 1024 * 1024;
cfg->sweep_count = 1024 * 16;
cfg->skip_start_factor = 0.0;
cfg->skip_end_factor = 1.0;
cfg->use_absolute_start_offset = false;
cfg->use_absolute_end_offset = false;
cfg->skip_start_offset = 0;
cfg->skip_end_offset = 0;
cfg->mtime_window = -1;
rm_trie_init(&cfg->file_trie);
}
static bool rm_tm_count_files(RmTrie *count_tree, char **paths, RmSession *session) {
if(*paths == NULL) {
rm_log_error("No paths passed to rm_tm_count_files\n");
return false;
}
int fts_flags = FTS_COMFOLLOW;
if(session->cfg->follow_symlinks) {
fts_flags |= FTS_LOGICAL;
} else {
fts_flags |= FTS_PHYSICAL;
}
/* This tree stores the full file paths.
It is joined into a full directory tree later.
*/
RmTrie file_tree;
rm_trie_init(&file_tree);
FTS *fts = fts_open(paths, fts_flags, NULL);
if(fts == NULL) {
rm_log_perror("fts_open failed");
return false;
}
FTSENT *ent = NULL;
while((ent = fts_read(fts))) {
/* Handle large files (where fts fails with FTS_NS) */
if(ent->fts_info == FTS_NS) {
RmStat stat_buf;
if(rm_sys_stat(ent->fts_path, &stat_buf) == -1) {
rm_log_perror("stat(2) failed");
continue;
} else {
/* Must be a large file (or followed link to it) */
ent->fts_info = FTS_F;
}
}
switch(ent->fts_info) {
case FTS_ERR:
case FTS_DC:
/* Save this path as an error */
rm_trie_insert(&file_tree, ent->fts_path, GINT_TO_POINTER(true));
break;
case FTS_F:
case FTS_SL:
case FTS_NS:
case FTS_SLNONE:
case FTS_DEFAULT:
/* Save this path as countable file */
if(ent->fts_statp->st_size > 0) {
rm_trie_insert(&file_tree, ent->fts_path, GINT_TO_POINTER(false));
}
case FTS_D:
case FTS_DNR:
case FTS_DOT:
case FTS_DP:
case FTS_NSOK:
default:
/* other fts states, that do not count as errors or files */
break;
}
}
if(fts_close(fts) != 0) {
rm_log_perror("fts_close failed");
return false;
}
rm_trie_iter(&file_tree, NULL, true, false, rm_tm_count_art_callback, count_tree);
/* Now flag everything as a no-go over the given paths,
* otherwise we would continue merging till / with fatal consequences,
* since / does not have more files as paths[0]
*/
for(int i = 0; paths[i]; ++i) {
/* Just call the callback directly */
RmNode *node = rm_trie_search_node(&file_tree, paths[i]);
if(node != NULL) {
node->data = GINT_TO_POINTER(true);
rm_tm_count_art_callback(&file_tree, node, 0, count_tree);
}
}
#ifdef _RM_TREEMERGE_DEBUG
rm_trie_print(count_tree);
#endif
rm_trie_destroy(&file_tree);
return true;
}