/* This does preprocessing including handling of "other lint" (non-dupes)
* After rm_preprocess(), all remaining duplicate candidates are in
* a jagged GSList of GSLists as follows:
* session->tables->size_groups->group1->file1a
* ->file1b
* ->file1c
* ->group2->file2a
* ->file2b
* etc
*/
void rm_preprocess(RmSession *session) {
RmFileTables *tables = session->tables;
GQueue *all_files = tables->all_files;
session->total_filtered_files = session->total_files;
/* initial sort by size */
g_queue_sort(all_files, (GCompareDataFunc)rm_file_cmp_full, session);
rm_log_debug_line("initial size sort finished at time %.3f; sorted %d files",
g_timer_elapsed(session->timer, NULL),
session->total_files);
/* split into file size groups; for each size, remove path doubles and bundle
* hardlinks */
rm_assert_gentle(all_files->head);
RmFile *file = g_queue_pop_head(all_files);
RmFile *current_size_file = file;
guint removed = 0;
GHashTable *node_table = tables->node_table;
while(file && !rm_session_was_aborted()) {
/* group files into inode clusters */
GQueue *inode_cluster =
rm_hash_table_setdefault(node_table, file, (RmNewFunc)g_queue_new);
g_queue_push_tail(inode_cluster, file);
/* get next file and check if it is part of the same group */
file = g_queue_pop_head(all_files);
if(!file || rm_file_cmp_split(file, current_size_file, session) != 0) {
/* process completed group (all same size & other criteria)*/
/* remove path doubles and handle "other" lint */
/* add an empty GSlist to our list of lists */
tables->size_groups = g_slist_prepend(tables->size_groups, NULL);
removed += g_hash_table_foreach_remove(
node_table, (GHRFunc)rm_pp_handle_inode_clusters, session);
/* free up the node table for the next group */
g_hash_table_steal_all(node_table);
if(tables->size_groups->data == NULL) {
/* zero size group after handling other lint; remove it */
tables->size_groups = g_slist_delete_link(tables->size_groups, tables->size_groups);
}
}
current_size_file = file;
}
session->other_lint_cnt += rm_pp_handler_other_lint(session);
rm_log_debug_line(
"path doubles removal/hardlink bundling/other lint finished at %.3f; removed %u "
"of %d",
g_timer_elapsed(session->timer, NULL), removed, session->total_files);
rm_fmt_set_state(session->formats, RM_PROGRESS_STATE_PREPROCESS);
}
static void rm_tm_insert_dir(RmTreeMerger *self, RmDirectory *directory) {
if(directory->was_inserted) {
return;
}
GQueue *dir_queue =
rm_hash_table_setdefault(self->result_table, directory, (RmNewFunc)g_queue_new);
g_queue_push_head(dir_queue, directory);
directory->was_inserted = true;
}
static void rm_tm_forward_unresolved(RmTreeMerger *self, RmDirectory *directory) {
if(directory->finished == true) {
return;
} else {
directory->finished = true;
}
for(GList *iter = directory->known_files.head; iter; iter = iter->next) {
RmFile *file = iter->data;
GQueue *file_list = rm_hash_table_setdefault(self->file_groups, file->digest,
(RmNewFunc)g_queue_new);
g_queue_push_head(file_list, file);
}
/* Recursively propagate to children */
for(GList *iter = directory->children.head; iter; iter = iter->next) {
rm_tm_forward_unresolved(self, (RmDirectory *)iter->data);
}
}
