2020-09-26 07:05:47 +12:00
use std ::collections ::BTreeMap ;
2023-05-03 08:37:12 +12:00
use std ::fs ;
2023-10-11 07:54:41 +13:00
use std ::fs ::{ DirEntry , Metadata } ;
use std ::io ::Write ;
2023-05-03 08:37:12 +12:00
use std ::path ::{ Path , PathBuf } ;
use std ::sync ::atomic ::{ AtomicUsize , Ordering } ;
2020-12-02 22:25:27 +13:00
use std ::sync ::Arc ;
2021-11-28 08:49:20 +13:00
use crossbeam_channel ::Receiver ;
2023-10-11 07:54:41 +13:00
use fun_time ::fun_time ;
2023-05-03 08:37:12 +12:00
use futures ::channel ::mpsc ::UnboundedSender ;
2023-07-10 18:36:03 +12:00
use humansize ::{ format_size , BINARY } ;
2023-10-11 07:54:41 +13:00
use log ::debug ;
2021-12-18 07:29:37 +13:00
use rayon ::prelude ::* ;
2023-10-13 05:48:46 +13:00
use serde ::{ Deserialize , Serialize } ;
2021-11-28 08:49:20 +13:00
2023-10-15 03:26:17 +13:00
use crate ::common ::{ check_folder_children , check_if_stop_received , prepare_thread_handler_common , send_info_and_wait_for_ending_all_threads , split_path } ;
2023-05-11 07:27:41 +12:00
use crate ::common_dir_traversal ::{ common_get_entry_data_metadata , common_read_dir , get_lowercase_name , get_modified_time , CheckingMethod , ProgressData , ToolType } ;
2023-10-11 07:54:41 +13:00
use crate ::common_tool ::{ CommonData , CommonToolData , DeleteMethod } ;
use crate ::common_traits ::{ DebugPrint , PrintResults } ;
2020-09-26 07:05:47 +12:00
2023-10-13 05:48:46 +13:00
#[ derive(Clone, Debug, Serialize, Deserialize) ]
2020-09-26 07:05:47 +12:00
pub struct FileEntry {
2020-10-15 05:41:37 +13:00
pub path : PathBuf ,
2020-09-26 07:05:47 +12:00
pub size : u64 ,
2020-10-01 17:53:10 +13:00
pub modified_date : u64 ,
2020-09-26 07:05:47 +12:00
}
2022-06-05 07:20:21 +12:00
#[ derive(Copy, Clone, Eq, PartialEq) ]
pub enum SearchMode {
BiggestFiles ,
SmallestFiles ,
}
2020-10-10 09:32:08 +13:00
#[ derive(Default) ]
2020-09-26 07:05:47 +12:00
pub struct Info {
2020-09-26 23:48:53 +12:00
pub number_of_real_files : usize ,
2020-09-26 07:05:47 +12:00
}
pub struct BigFile {
2023-10-05 19:06:47 +13:00
common_data : CommonToolData ,
2020-09-26 07:05:47 +12:00
information : Info ,
2023-10-13 05:48:46 +13:00
big_files : Vec < FileEntry > ,
2020-09-26 07:05:47 +12:00
number_of_files_to_check : usize ,
2022-06-05 07:20:21 +12:00
search_mode : SearchMode ,
2020-09-26 07:05:47 +12:00
}
impl BigFile {
2020-10-10 09:32:08 +13:00
pub fn new ( ) -> Self {
Self {
2023-10-05 19:06:47 +13:00
common_data : CommonToolData ::new ( ToolType ::BigFile ) ,
information : Info ::default ( ) ,
2020-09-26 07:05:47 +12:00
big_files : Default ::default ( ) ,
number_of_files_to_check : 50 ,
2022-07-29 04:56:55 +12:00
search_mode : SearchMode ::BiggestFiles ,
2020-09-26 07:05:47 +12:00
}
}
2020-09-27 06:51:28 +13:00
2023-10-15 02:04:45 +13:00
#[ fun_time(message = " find_big_files " , level = " info " ) ]
2023-05-03 08:37:12 +12:00
pub fn find_big_files ( & mut self , stop_receiver : Option < & Receiver < ( ) > > , progress_sender : Option < & UnboundedSender < ProgressData > > ) {
2023-10-05 19:06:47 +13:00
self . optimize_dirs_before_start ( ) ;
2020-12-02 22:25:27 +13:00
if ! self . look_for_big_files ( stop_receiver , progress_sender ) {
2023-10-05 19:06:47 +13:00
self . common_data . stopped_search = true ;
2020-10-10 22:15:20 +13:00
return ;
}
2020-11-01 07:19:33 +13:00
self . delete_files ( ) ;
2020-09-27 06:51:28 +13:00
self . debug_print ( ) ;
}
2020-09-26 07:05:47 +12:00
2023-10-15 02:04:45 +13:00
#[ fun_time(message = " look_for_big_files " , level = " debug " ) ]
2023-05-03 08:37:12 +12:00
fn look_for_big_files ( & mut self , stop_receiver : Option < & Receiver < ( ) > > , progress_sender : Option < & UnboundedSender < ProgressData > > ) -> bool {
2020-10-15 05:41:37 +13:00
let mut folders_to_check : Vec < PathBuf > = Vec ::with_capacity ( 1024 * 2 ) ; // This should be small enough too not see to big difference and big enough to store most of paths without needing to resize vector
2022-06-05 07:20:21 +12:00
let mut old_map : BTreeMap < u64 , Vec < FileEntry > > = Default ::default ( ) ;
2020-09-26 07:05:47 +12:00
// Add root folders for finding
2023-10-05 19:06:47 +13:00
for id in & self . common_data . directories . included_directories {
2020-10-15 05:41:37 +13:00
folders_to_check . push ( id . clone ( ) ) ;
2020-09-26 07:05:47 +12:00
}
2023-05-11 07:27:41 +12:00
let ( progress_thread_handle , progress_thread_run , atomic_counter , _check_was_stopped ) =
2023-10-05 19:06:47 +13:00
prepare_thread_handler_common ( progress_sender , 0 , 0 , 0 , CheckingMethod ::None , self . common_data . tool_type ) ;
2020-12-02 22:25:27 +13:00
2023-10-11 07:54:41 +13:00
debug! ( " Starting to search for big files " ) ;
2020-09-26 07:05:47 +12:00
while ! folders_to_check . is_empty ( ) {
2023-10-15 03:26:17 +13:00
if check_if_stop_received ( stop_receiver ) {
2023-05-03 08:37:12 +12:00
send_info_and_wait_for_ending_all_threads ( & progress_thread_run , progress_thread_handle ) ;
2020-10-10 22:15:20 +13:00
return false ;
}
2020-12-02 22:25:27 +13:00
2021-12-18 07:29:37 +13:00
let segments : Vec < _ > = folders_to_check
. par_iter ( )
. map ( | current_folder | {
let mut dir_result = vec! [ ] ;
let mut warnings = vec! [ ] ;
let mut fe_result = vec! [ ] ;
2023-05-03 08:37:12 +12:00
let Some ( read_dir ) = common_read_dir ( current_folder , & mut warnings ) else {
return ( dir_result , warnings , fe_result ) ;
2021-12-18 07:29:37 +13:00
} ;
2020-09-26 07:05:47 +12:00
2021-12-18 07:29:37 +13:00
// Check every sub folder/file/link etc.
2023-05-03 08:37:12 +12:00
for entry in read_dir {
let Some ( ( entry_data , metadata ) ) = common_get_entry_data_metadata ( & entry , & mut warnings , current_folder ) else {
continue ;
2021-12-18 07:29:37 +13:00
} ;
2023-05-03 08:37:12 +12:00
2021-12-18 07:29:37 +13:00
if metadata . is_dir ( ) {
2023-05-03 08:37:12 +12:00
check_folder_children (
& mut dir_result ,
& mut warnings ,
current_folder ,
entry_data ,
2023-10-05 19:06:47 +13:00
self . common_data . recursive_search ,
& self . common_data . directories ,
& self . common_data . excluded_items ,
2023-05-03 08:37:12 +12:00
) ;
2021-12-18 07:29:37 +13:00
} else if metadata . is_file ( ) {
2023-05-03 08:37:12 +12:00
self . collect_file_entry ( & atomic_counter , & metadata , entry_data , & mut fe_result , & mut warnings , current_folder ) ;
2021-12-18 07:29:37 +13:00
}
}
( dir_result , warnings , fe_result )
} )
. collect ( ) ;
// Advance the frontier
folders_to_check . clear ( ) ;
// Process collected data
for ( segment , warnings , fe_result ) in segments {
folders_to_check . extend ( segment ) ;
2023-10-05 19:06:47 +13:00
self . common_data . text_messages . warnings . extend ( warnings ) ;
2021-12-18 07:29:37 +13:00
for ( size , fe ) in fe_result {
2023-10-05 19:06:47 +13:00
old_map . entry ( size ) . or_default ( ) . push ( fe ) ;
2020-09-26 07:05:47 +12:00
}
}
}
2023-10-11 07:54:41 +13:00
debug! ( " Collected {} files " , old_map . len ( ) ) ;
2023-05-03 08:37:12 +12:00
send_info_and_wait_for_ending_all_threads ( & progress_thread_run , progress_thread_handle ) ;
self . extract_n_biggest_files ( old_map ) ;
true
}
2020-12-02 22:25:27 +13:00
2023-05-03 08:37:12 +12:00
pub fn collect_file_entry (
& self ,
atomic_counter : & Arc < AtomicUsize > ,
metadata : & Metadata ,
entry_data : & DirEntry ,
fe_result : & mut Vec < ( u64 , FileEntry ) > ,
warnings : & mut Vec < String > ,
current_folder : & Path ,
) {
atomic_counter . fetch_add ( 1 , Ordering ::Relaxed ) ;
if metadata . len ( ) = = 0 {
return ;
}
2020-09-26 07:05:47 +12:00
2023-05-03 08:37:12 +12:00
let Some ( file_name_lowercase ) = get_lowercase_name ( entry_data , warnings ) else {
return ;
} ;
2023-10-05 19:06:47 +13:00
if ! self . common_data . allowed_extensions . matches_filename ( & file_name_lowercase ) {
2023-05-03 08:37:12 +12:00
return ;
}
let current_file_name = current_folder . join ( entry_data . file_name ( ) ) ;
2023-10-05 19:06:47 +13:00
if self . common_data . excluded_items . is_excluded ( & current_file_name ) {
2023-05-03 08:37:12 +12:00
return ;
}
let fe : FileEntry = FileEntry {
path : current_file_name . clone ( ) ,
size : metadata . len ( ) ,
modified_date : get_modified_time ( metadata , warnings , & current_file_name , false ) ,
} ;
fe_result . push ( ( fe . size , fe ) ) ;
}
2023-10-15 02:04:45 +13:00
#[ fun_time(message = " extract_n_biggest_files " , level = " debug " ) ]
2023-05-03 08:37:12 +12:00
pub fn extract_n_biggest_files ( & mut self , old_map : BTreeMap < u64 , Vec < FileEntry > > ) {
2022-06-05 07:20:21 +12:00
let iter : Box < dyn Iterator < Item = _ > > ;
if self . search_mode = = SearchMode ::SmallestFiles {
iter = Box ::new ( old_map . into_iter ( ) ) ;
} else {
iter = Box ::new ( old_map . into_iter ( ) . rev ( ) ) ;
}
2023-10-13 05:48:46 +13:00
for ( _size , mut vector ) in iter {
2020-09-26 23:48:53 +12:00
if self . information . number_of_real_files < self . number_of_files_to_check {
2022-06-05 07:20:21 +12:00
if vector . len ( ) > 1 {
2022-07-25 06:48:02 +12:00
vector . sort_unstable_by_key ( | e | {
2022-06-05 07:20:21 +12:00
let t = split_path ( e . path . as_path ( ) ) ;
( t . 0 , t . 1 )
} ) ;
}
2020-09-26 07:05:47 +12:00
for file in vector {
2020-09-26 23:48:53 +12:00
if self . information . number_of_real_files < self . number_of_files_to_check {
2023-10-13 05:48:46 +13:00
self . big_files . push ( file ) ;
2020-09-26 23:48:53 +12:00
self . information . number_of_real_files + = 1 ;
2020-09-26 07:05:47 +12:00
} else {
break ;
}
}
} else {
break ;
}
}
2020-09-27 03:52:13 +13:00
}
2020-11-01 07:19:33 +13:00
fn delete_files ( & mut self ) {
2023-10-11 07:54:41 +13:00
match self . common_data . delete_method {
2020-11-01 07:19:33 +13:00
DeleteMethod ::Delete = > {
2023-10-13 05:48:46 +13:00
for file_entry in & self . big_files {
2023-02-19 22:21:14 +13:00
if fs ::remove_file ( & file_entry . path ) . is_err ( ) {
2023-10-05 19:06:47 +13:00
self . common_data . text_messages . warnings . push ( file_entry . path . display ( ) . to_string ( ) ) ;
2020-11-01 07:19:33 +13:00
}
}
}
DeleteMethod ::None = > {
//Just do nothing
}
2023-10-11 07:54:41 +13:00
_ = > unreachable! ( ) ,
2020-11-01 07:19:33 +13:00
}
}
2020-09-27 03:52:13 +13:00
}
impl Default for BigFile {
fn default ( ) -> Self {
Self ::new ( )
}
}
impl DebugPrint for BigFile {
2020-09-26 07:05:47 +12:00
fn debug_print ( & self ) {
2023-10-11 07:54:41 +13:00
if ! cfg! ( debug_assertions ) {
2020-09-26 07:05:47 +12:00
return ;
}
2020-09-26 23:48:53 +12:00
2023-10-05 19:06:47 +13:00
println! ( " ### INDIVIDUAL DEBUG PRINT ### " ) ;
2020-09-26 23:48:53 +12:00
println! ( " Big files size {} in {} groups " , self . information . number_of_real_files , self . big_files . len ( ) ) ;
println! ( " Number of files to check - {:?} " , self . number_of_files_to_check ) ;
2023-10-05 19:06:47 +13:00
self . debug_print_common ( ) ;
2020-09-26 07:05:47 +12:00
println! ( " ----------------------------------------- " ) ;
}
2020-09-27 03:52:13 +13:00
}
2020-10-15 05:41:37 +13:00
2023-10-11 07:54:41 +13:00
impl PrintResults for BigFile {
fn write_results < T : Write > ( & self , writer : & mut T ) -> std ::io ::Result < ( ) > {
writeln! (
2021-01-09 23:52:43 +13:00
writer ,
2020-10-15 05:41:37 +13:00
" Results of searching {:?} with excluded directories {:?} and excluded items {:?} " ,
2023-10-05 19:06:47 +13:00
self . common_data . directories . included_directories , self . common_data . directories . excluded_directories , self . common_data . excluded_items . items
2023-10-11 07:54:41 +13:00
) ? ;
2020-09-26 07:05:47 +12:00
2020-09-27 03:52:13 +13:00
if self . information . number_of_real_files ! = 0 {
2022-07-31 18:45:19 +12:00
if self . search_mode = = SearchMode ::BiggestFiles {
2023-10-11 07:54:41 +13:00
writeln! ( writer , " {} the biggest files. \n \n " , self . information . number_of_real_files ) ? ;
2022-07-31 18:45:19 +12:00
} else {
2023-10-11 07:54:41 +13:00
writeln! ( writer , " {} the smallest files. \n \n " , self . information . number_of_real_files ) ? ;
2022-07-31 18:45:19 +12:00
}
2023-10-13 05:48:46 +13:00
for file_entry in & self . big_files {
writeln! ( writer , " {} ({}) - {} " , format_size ( file_entry . size , BINARY ) , file_entry . size , file_entry . path . display ( ) ) ? ;
2020-09-26 07:05:47 +12:00
}
2020-09-27 03:52:13 +13:00
} else {
2021-01-09 23:52:43 +13:00
write! ( writer , " Not found any files. " ) . unwrap ( ) ;
2020-09-26 07:05:47 +12:00
}
2023-05-08 06:54:05 +12:00
2023-10-11 07:54:41 +13:00
Ok ( ( ) )
2020-09-26 07:05:47 +12:00
}
2023-10-13 05:48:46 +13:00
fn save_results_to_file_as_json ( & self , file_name : & str , pretty_print : bool ) -> std ::io ::Result < ( ) > {
self . save_results_to_file_as_json_internal ( file_name , & self . big_files , pretty_print )
}
2020-09-26 07:05:47 +12:00
}
2023-10-05 19:06:47 +13:00
impl CommonData for BigFile {
fn get_cd ( & self ) -> & CommonToolData {
& self . common_data
}
fn get_cd_mut ( & mut self ) -> & mut CommonToolData {
& mut self . common_data
}
}
impl BigFile {
pub fn set_search_mode ( & mut self , search_mode : SearchMode ) {
self . search_mode = search_mode ;
}
2023-10-13 05:48:46 +13:00
pub const fn get_big_files ( & self ) -> & Vec < FileEntry > {
2023-10-05 19:06:47 +13:00
& self . big_files
}
pub const fn get_information ( & self ) -> & Info {
& self . information
}
pub fn set_number_of_files_to_check ( & mut self , number_of_files_to_check : usize ) {
self . number_of_files_to_check = number_of_files_to_check ;
}
}