2020-09-27 03:52:13 +13:00
use crate ::common ::Common ;
use crate ::common_directory ::Directories ;
use crate ::common_extensions ::Extensions ;
use crate ::common_items ::ExcludedItems ;
use crate ::common_messages ::Messages ;
use crate ::common_traits ::{ DebugPrint , PrintResults , SaveResults } ;
2020-10-10 22:15:20 +13:00
use crossbeam_channel ::Receiver ;
2020-09-26 23:48:53 +12:00
use humansize ::{ file_size_opts as options , FileSize } ;
2020-09-26 07:05:47 +12:00
use std ::collections ::BTreeMap ;
2020-10-15 05:41:37 +13:00
use std ::ffi ::OsStr ;
2020-09-26 07:05:47 +12:00
use std ::fs ;
2020-09-26 23:48:53 +12:00
use std ::fs ::{ File , Metadata } ;
use std ::io ::Write ;
2020-10-15 05:41:37 +13:00
use std ::path ::PathBuf ;
2020-10-01 17:53:10 +13:00
use std ::time ::{ SystemTime , UNIX_EPOCH } ;
2020-09-26 07:05:47 +12:00
#[ derive(Clone) ]
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
}
2020-11-01 07:19:33 +13:00
#[ derive(Eq, PartialEq, Clone, Debug) ]
pub enum DeleteMethod {
None ,
Delete ,
}
2020-09-26 07:05:47 +12:00
/// Info struck with helpful information's about results
2020-10-10 09:32:08 +13:00
#[ derive(Default) ]
2020-09-26 07:05:47 +12:00
pub struct Info {
pub number_of_checked_files : usize ,
pub number_of_checked_folders : usize ,
pub number_of_ignored_files : usize ,
pub number_of_ignored_things : usize ,
pub taken_space : u64 ,
2020-09-26 23:48:53 +12:00
pub number_of_real_files : usize ,
2020-09-26 07:05:47 +12:00
}
2020-10-10 09:32:08 +13:00
impl Info {
pub fn new ( ) -> Self {
Default ::default ( )
2020-09-26 07:05:47 +12:00
}
}
/// Struct with required information's to work
pub struct BigFile {
text_messages : Messages ,
information : Info ,
big_files : BTreeMap < u64 , Vec < FileEntry > > ,
2020-09-27 03:52:13 +13:00
excluded_items : ExcludedItems ,
directories : Directories ,
allowed_extensions : Extensions ,
2020-09-26 07:05:47 +12:00
recursive_search : bool ,
number_of_files_to_check : usize ,
2020-11-01 07:19:33 +13:00
delete_method : DeleteMethod ,
2020-10-10 22:15:20 +13:00
stopped_search : bool ,
2020-09-26 07:05:47 +12:00
}
impl BigFile {
2020-10-10 09:32:08 +13:00
pub fn new ( ) -> Self {
Self {
2020-09-26 07:05:47 +12:00
text_messages : Default ::default ( ) ,
information : Info ::new ( ) ,
big_files : Default ::default ( ) ,
2020-09-27 03:52:13 +13:00
excluded_items : ExcludedItems ::new ( ) ,
directories : Directories ::new ( ) ,
allowed_extensions : Extensions ::new ( ) ,
2020-09-26 23:48:53 +12:00
recursive_search : true ,
2020-09-26 07:05:47 +12:00
number_of_files_to_check : 50 ,
2020-11-01 07:19:33 +13:00
delete_method : DeleteMethod ::None ,
2020-10-10 22:15:20 +13:00
stopped_search : false ,
2020-09-26 07:05:47 +12:00
}
}
2020-09-27 06:51:28 +13:00
2020-11-05 06:15:06 +13:00
pub fn find_big_files ( & mut self , stop_receiver : Option < & Receiver < ( ) > > ) {
2020-09-27 06:51:28 +13:00
self . optimize_directories ( ) ;
2020-11-05 06:15:06 +13:00
if ! self . look_for_big_files ( stop_receiver ) {
2020-10-10 22:15:20 +13:00
self . stopped_search = true ;
return ;
}
2020-11-01 07:19:33 +13:00
self . delete_files ( ) ;
2020-09-27 06:51:28 +13:00
self . debug_print ( ) ;
}
2020-10-10 22:15:20 +13:00
pub fn get_stopped_search ( & self ) -> bool {
self . stopped_search
}
2020-09-27 06:51:28 +13:00
2020-10-10 09:32:08 +13:00
pub const fn get_big_files ( & self ) -> & BTreeMap < u64 , Vec < FileEntry > > {
2020-10-02 06:26:25 +13:00
& self . big_files
}
2020-10-10 09:32:08 +13:00
pub const fn get_text_messages ( & self ) -> & Messages {
2020-09-26 07:05:47 +12:00
& self . text_messages
}
2020-10-10 09:32:08 +13:00
pub const fn get_information ( & self ) -> & Info {
2020-09-26 07:05:47 +12:00
& self . information
}
2020-11-01 07:19:33 +13:00
pub fn set_delete_method ( & mut self , delete_method : DeleteMethod ) {
self . delete_method = delete_method ;
}
2020-09-26 23:48:53 +12:00
pub fn set_recursive_search ( & mut self , recursive_search : bool ) {
self . recursive_search = recursive_search ;
}
2020-09-26 07:05:47 +12:00
/// List of allowed extensions, only files with this extensions will be checking if are duplicates
2020-09-27 03:52:13 +13:00
pub fn set_allowed_extensions ( & mut self , allowed_extensions : String ) {
self . allowed_extensions . set_allowed_extensions ( allowed_extensions , & mut self . text_messages ) ;
2020-09-26 07:05:47 +12:00
}
2020-11-05 06:15:06 +13:00
fn look_for_big_files ( & mut self , stop_receiver : Option < & Receiver < ( ) > > ) -> bool {
2020-09-26 07:05:47 +12:00
let start_time : SystemTime = SystemTime ::now ( ) ;
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
2020-09-26 07:05:47 +12:00
// Add root folders for finding
2020-09-27 03:52:13 +13:00
for id in & self . 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
}
self . information . number_of_checked_folders + = folders_to_check . len ( ) ;
while ! folders_to_check . is_empty ( ) {
2020-11-05 06:15:06 +13:00
if stop_receiver . is_some ( ) & & stop_receiver . unwrap ( ) . try_recv ( ) . is_ok ( ) {
2020-10-10 22:15:20 +13:00
return false ;
}
2020-10-15 05:41:37 +13:00
let current_folder = folders_to_check . pop ( ) . unwrap ( ) ;
2020-09-26 07:05:47 +12:00
let read_dir = match fs ::read_dir ( & current_folder ) {
Ok ( t ) = > t ,
Err ( _ ) = > {
2020-10-15 05:41:37 +13:00
self . text_messages . warnings . push ( format! ( " Cannot open dir {} " , current_folder . display ( ) ) ) ;
2020-09-26 07:05:47 +12:00
continue ;
} // Permissions denied
} ;
2020-10-10 09:32:08 +13:00
' dir : for entry in read_dir {
2020-09-26 07:05:47 +12:00
let entry_data = match entry {
Ok ( t ) = > t ,
Err ( _ ) = > {
2020-10-15 05:41:37 +13:00
self . text_messages . warnings . push ( format! ( " Cannot read entry in dir {} " , current_folder . display ( ) ) ) ;
2020-09-26 07:05:47 +12:00
continue ;
} //Permissions denied
} ;
let metadata : Metadata = match entry_data . metadata ( ) {
Ok ( t ) = > t ,
Err ( _ ) = > {
2020-10-15 05:41:37 +13:00
self . text_messages . warnings . push ( format! ( " Cannot read metadata in dir {} " , current_folder . display ( ) ) ) ;
2020-09-26 07:05:47 +12:00
continue ;
} //Permissions denied
} ;
if metadata . is_dir ( ) {
self . information . number_of_checked_folders + = 1 ;
if ! self . recursive_search {
continue ;
}
2020-10-15 05:41:37 +13:00
let next_folder = current_folder . join ( entry_data . file_name ( ) ) ;
if self . directories . is_excluded ( & next_folder ) | | self . excluded_items . is_excluded ( & next_folder ) {
continue 'dir ;
2020-09-26 07:05:47 +12:00
}
2020-10-15 05:41:37 +13:00
2020-10-10 09:32:08 +13:00
folders_to_check . push ( next_folder ) ;
2020-09-26 07:05:47 +12:00
} else if metadata . is_file ( ) {
2020-10-15 05:41:37 +13:00
// Extracting file extension
let file_extension = entry_data . path ( ) . extension ( ) . and_then ( OsStr ::to_str ) . map ( str ::to_lowercase ) ;
2020-09-26 07:05:47 +12:00
// Checking allowed extensions
2020-10-10 22:15:20 +13:00
if ! self . allowed_extensions . file_extensions . is_empty ( ) {
2020-10-15 05:41:37 +13:00
let allowed = self . allowed_extensions . file_extensions . iter ( ) . map ( | e | e . to_lowercase ( ) ) . any ( | e | file_extension = = Some ( e ) ) ;
2020-10-10 22:15:20 +13:00
if ! allowed {
// Not an allowed extension, ignore it.
self . information . number_of_ignored_files + = 1 ;
continue 'dir ;
}
2020-09-26 07:05:47 +12:00
}
2020-10-10 09:32:08 +13:00
// Checking expressions
2020-10-15 05:41:37 +13:00
let current_file_name = current_folder . join ( entry_data . file_name ( ) ) ;
if self . excluded_items . is_excluded ( & current_file_name ) {
continue 'dir ;
2020-10-11 02:18:04 +13:00
}
2020-10-10 09:32:08 +13:00
// Creating new file entry
let fe : FileEntry = FileEntry {
path : current_file_name . clone ( ) ,
size : metadata . len ( ) ,
modified_date : match metadata . modified ( ) {
Ok ( t ) = > match t . duration_since ( UNIX_EPOCH ) {
Ok ( d ) = > d . as_secs ( ) ,
2020-09-26 07:05:47 +12:00
Err ( _ ) = > {
2020-10-15 05:41:37 +13:00
self . text_messages . warnings . push ( format! ( " File {} seems to be modified before Unix Epoch. " , current_file_name . display ( ) ) ) ;
2020-10-10 09:32:08 +13:00
0
}
2020-09-26 07:05:47 +12:00
} ,
2020-10-10 09:32:08 +13:00
Err ( _ ) = > {
2020-10-15 05:41:37 +13:00
self . text_messages . warnings . push ( format! ( " Unable to get modification date from file {} " , current_file_name . display ( ) ) ) ;
2020-10-10 09:32:08 +13:00
continue ;
} // Permissions Denied
} ,
} ;
2020-09-26 07:05:47 +12:00
2020-10-10 09:32:08 +13:00
self . big_files . entry ( metadata . len ( ) ) . or_insert_with ( Vec ::new ) ;
self . big_files . get_mut ( & metadata . len ( ) ) . unwrap ( ) . push ( fe ) ;
2020-09-26 07:05:47 +12:00
2020-10-10 09:32:08 +13:00
self . information . number_of_checked_files + = 1 ;
2020-09-26 07:05:47 +12:00
} else {
// Probably this is symbolic links so we are free to ignore this
self . information . number_of_ignored_things + = 1 ;
}
}
}
2020-09-27 06:51:28 +13:00
// Extract n biggest files to new TreeMap
2020-09-26 07:05:47 +12:00
let mut new_map : BTreeMap < u64 , Vec < FileEntry > > = Default ::default ( ) ;
2020-09-26 23:48:53 +12:00
for ( size , vector ) in self . big_files . iter ( ) . rev ( ) {
if self . information . number_of_real_files < self . number_of_files_to_check {
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 {
2020-09-26 07:05:47 +12:00
new_map . entry ( * size ) . or_insert_with ( Vec ::new ) ;
new_map . get_mut ( size ) . unwrap ( ) . push ( file . clone ( ) ) ;
2020-09-26 23:48:53 +12:00
self . information . taken_space + = size ;
self . information . number_of_real_files + = 1 ;
2020-09-26 07:05:47 +12:00
} else {
break ;
}
}
} else {
break ;
}
}
self . big_files = new_map ;
Common ::print_time ( start_time , SystemTime ::now ( ) , " look_for_big_files " . to_string ( ) ) ;
2020-10-10 22:15:20 +13:00
true
2020-09-26 07:05:47 +12:00
}
2020-09-27 03:52:13 +13:00
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 ;
}
2020-10-07 21:34:15 +13:00
/// Setting excluded items which needs to contains * wildcard
2020-09-27 03:52:13 +13:00
/// Are a lot of slower than absolute path, so it should be used to heavy
pub fn set_excluded_items ( & mut self , excluded_items : String ) {
self . excluded_items . set_excluded_items ( excluded_items , & mut self . text_messages ) ;
}
/// Remove unused entries when included or excluded overlaps with each other or are duplicated etc.
fn optimize_directories ( & mut self ) {
self . directories . optimize_directories ( self . recursive_search , & mut self . text_messages ) ;
}
2020-09-27 04:16:12 +13:00
/// Setting included directories, at least one must be provided
2020-09-27 03:52:13 +13:00
pub fn set_included_directory ( & mut self , included_directory : String ) {
self . directories . set_included_directory ( included_directory , & mut self . text_messages ) ;
}
/// Setting absolute path to exclude
pub fn set_excluded_directory ( & mut self , excluded_directory : String ) {
self . directories . set_excluded_directory ( excluded_directory , & mut self . text_messages ) ;
}
2020-11-01 07:19:33 +13:00
/// Function to delete files, from filed Vector
fn delete_files ( & mut self ) {
let start_time : SystemTime = SystemTime ::now ( ) ;
match self . delete_method {
DeleteMethod ::Delete = > {
for vec_file_entry in self . big_files . values ( ) {
for file_entry in vec_file_entry {
if fs ::remove_file ( file_entry . path . clone ( ) ) . is_err ( ) {
self . text_messages . warnings . push ( file_entry . path . display ( ) . to_string ( ) ) ;
}
}
}
}
DeleteMethod ::None = > {
//Just do nothing
}
}
Common ::print_time ( start_time , SystemTime ::now ( ) , " delete_files " . to_string ( ) ) ;
}
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
#[ allow(dead_code) ]
#[ allow(unreachable_code) ]
2020-09-26 23:48:53 +12:00
/// Debugging printing - only available on debug build
2020-09-26 07:05:47 +12:00
fn debug_print ( & self ) {
#[ cfg(not(debug_assertions)) ]
{
return ;
}
println! ( " ---------------DEBUG PRINT--------------- " ) ;
2020-09-26 23:48:53 +12:00
println! ( " ### Information's " ) ;
println! ( " Errors size - {} " , self . text_messages . errors . len ( ) ) ;
println! ( " Warnings size - {} " , self . text_messages . warnings . len ( ) ) ;
println! ( " Messages size - {} " , self . text_messages . messages . len ( ) ) ;
println! ( " Number of checked files - {} " , self . information . number_of_checked_files ) ;
println! ( " Number of checked folders - {} " , self . information . number_of_checked_folders ) ;
println! ( " Number of ignored files - {} " , self . information . number_of_ignored_files ) ;
println! ( " Number of ignored things(like symbolic links) - {} " , self . information . number_of_ignored_things ) ;
println! ( " ### Other " ) ;
println! ( " Big files size {} in {} groups " , self . information . number_of_real_files , self . big_files . len ( ) ) ;
2020-09-27 03:52:13 +13:00
println! ( " Allowed extensions - {:?} " , self . allowed_extensions . file_extensions ) ;
println! ( " Excluded items - {:?} " , self . excluded_items . items ) ;
println! ( " Included directories - {:?} " , self . directories . included_directories ) ;
println! ( " Excluded directories - {:?} " , self . directories . excluded_directories ) ;
2020-09-26 23:48:53 +12:00
println! ( " Recursive search - {} " , self . recursive_search . to_string ( ) ) ;
println! ( " Number of files to check - {:?} " , self . number_of_files_to_check ) ;
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
2020-09-27 03:52:13 +13:00
impl SaveResults for BigFile {
/// Saving results to provided file
fn save_results_to_file ( & mut self , file_name : & str ) -> bool {
2020-09-26 07:05:47 +12:00
let start_time : SystemTime = SystemTime ::now ( ) ;
2020-09-27 03:52:13 +13:00
let file_name : String = match file_name {
" " = > " results.txt " . to_string ( ) ,
k = > k . to_string ( ) ,
} ;
2020-09-26 07:05:47 +12:00
2020-09-27 03:52:13 +13:00
let mut file = match File ::create ( & file_name ) {
Ok ( t ) = > t ,
Err ( _ ) = > {
self . text_messages . errors . push ( " Failed to create file " . to_string ( ) + file_name . as_str ( ) ) ;
return false ;
2020-09-26 07:05:47 +12:00
}
2020-09-27 03:52:13 +13:00
} ;
2020-09-26 07:05:47 +12:00
2020-10-15 05:41:37 +13:00
if writeln! (
file ,
" Results of searching {:?} with excluded directories {:?} and excluded items {:?} " ,
self . directories . included_directories , self . directories . excluded_directories , self . excluded_items . items
)
. is_err ( )
{
self . text_messages . errors . push ( format! ( " Failed to save results to file {} " , file_name ) ) ;
return false ;
2020-09-26 07:05:47 +12:00
}
2020-09-27 03:52:13 +13:00
if self . information . number_of_real_files ! = 0 {
2020-10-15 05:41:37 +13:00
write! ( file , " {} the biggest files. \n \n " , self . information . number_of_real_files ) . unwrap ( ) ;
2020-09-26 07:05:47 +12:00
2020-09-27 03:52:13 +13:00
for ( size , files ) in self . big_files . iter ( ) . rev ( ) {
for file_entry in files {
2020-10-15 05:41:37 +13:00
writeln! ( file , " {} ({}) - {} " , size . file_size ( options ::BINARY ) . unwrap ( ) , size , file_entry . path . display ( ) ) . unwrap ( ) ;
2020-09-26 07:05:47 +12:00
}
}
2020-09-27 03:52:13 +13:00
} else {
2020-10-15 08:10:27 +13:00
write! ( file , " Not found any files. " ) . unwrap ( ) ;
2020-09-26 07:05:47 +12:00
}
2020-09-27 03:52:13 +13:00
Common ::print_time ( start_time , SystemTime ::now ( ) , " save_results_to_file " . to_string ( ) ) ;
2020-09-26 07:05:47 +12:00
true
}
2020-09-27 03:52:13 +13:00
}
2020-10-15 05:41:37 +13:00
2020-09-27 03:52:13 +13:00
impl PrintResults for BigFile {
fn print_results ( & self ) {
2020-09-26 07:05:47 +12:00
let start_time : SystemTime = SystemTime ::now ( ) ;
2020-09-27 03:52:13 +13:00
println! ( " Found {} files which take {} : " , self . information . number_of_real_files , self . information . taken_space . file_size ( options ::BINARY ) . unwrap ( ) ) ;
for ( size , vector ) in self . big_files . iter ( ) . rev ( ) {
// TODO Align all to same width
for entry in vector {
2020-10-15 05:41:37 +13:00
println! ( " {} ( {} bytes) - {} " , size . file_size ( options ::BINARY ) . unwrap ( ) , size , entry . path . display ( ) ) ;
2020-09-26 07:05:47 +12:00
}
}
2020-09-27 08:50:16 +13:00
Common ::print_time ( start_time , SystemTime ::now ( ) , " print_entries " . to_string ( ) ) ;
2020-09-26 07:05:47 +12:00
}
}