gint rm_file_cmp_full(const RmFile *file_a, const RmFile *file_b,
const RmSession *session) {
gint result = rm_file_cmp(file_a, file_b);
if(result != 0) {
return result;
}
return rm_pp_cmp_orig_criteria(file_a, file_b, session);
}
/* 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(session)) {
/* 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(file, current_size_file) != 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 gint rm_file_cmp_full(const RmFile *file_a, const RmFile *file_b,
const RmSession *session) {
gint result = rm_file_cmp(file_a, file_b);
if(result != 0) {
return result;
}
if(session->cfg->mtime_window >= 0) {
gdouble diff = file_a->mtime - file_b->mtime;
if(FLOAT_IS_ZERO(diff)) {
return 0;
}
return (diff < 0) ? -1 : +1;
}
return rm_pp_cmp_orig_criteria(file_a, file_b, session);
}
static gint rm_file_cmp_split(const RmFile *file_a, const RmFile *file_b,
const RmSession *session) {
gint result = rm_file_cmp(file_a, file_b);
if(result != 0) {
return result;
}
/* If --mtime-window is specified, we need to check if the mtime is inside
* the window. The file list was sorted by rm_file_cmp_full by taking the
* diff of mtimes, therefore we have to define the split criteria
* differently.
*/
if(session->cfg->mtime_window >= 0) {
gdouble diff = file_a->mtime - file_b->mtime;
if(FLOAT_IS_ZERO(diff - session->cfg->mtime_window) || fabs(diff) < session->cfg->mtime_window) {
return 0;
}
/* Split the group. */
return (diff < 0) ? -1 : +1;
}
return 0;
}
