1
0
Fork 0
mirror of synced 2024-04-29 01:52:39 +12:00
czkawka/czkawka_core/src/empty_files.rs

400 lines
15 KiB
Rust
Raw Normal View History

use rayon::prelude::*;
2020-09-27 08:50:16 +13:00
use std::fs::{File, Metadata};
use std::io::prelude::*;
2021-11-28 08:49:20 +13:00
use std::io::BufWriter;
use std::path::PathBuf;
2021-11-28 08:49:20 +13:00
use std::sync::atomic::{AtomicBool, AtomicUsize, Ordering};
use std::sync::Arc;
use std::thread::sleep;
2020-12-02 22:25:27 +13:00
use std::time::{Duration, SystemTime, UNIX_EPOCH};
use std::{fs, thread};
2020-09-27 08:50:16 +13:00
2021-11-28 08:49:20 +13:00
use crossbeam_channel::Receiver;
2020-09-27 08:50:16 +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::*;
2020-12-02 22:25:27 +13:00
#[derive(Debug)]
pub struct ProgressData {
pub current_stage: u8,
pub max_stage: u8,
pub files_checked: usize,
}
2020-09-27 08:50:16 +13:00
#[derive(Eq, PartialEq, Clone, Debug)]
pub enum DeleteMethod {
None,
Delete,
}
#[derive(Clone)]
pub struct FileEntry {
pub path: PathBuf,
2020-10-01 17:53:10 +13:00
pub modified_date: u64,
2020-09-27 08:50:16 +13:00
}
/// Info struck with helpful information's about results
#[derive(Default)]
2020-09-27 08:50:16 +13:00
pub struct Info {
pub number_of_empty_files: usize,
pub number_of_removed_files: usize,
pub number_of_failed_to_remove_files: usize,
}
2021-11-28 08:57:10 +13:00
2020-09-27 08:50:16 +13:00
impl Info {
pub fn new() -> Self {
Default::default()
2020-09-27 08:50:16 +13:00
}
}
/// Struct with required information's to work
pub struct EmptyFiles {
text_messages: Messages,
information: Info,
empty_files: Vec<FileEntry>,
directories: Directories,
allowed_extensions: Extensions,
excluded_items: ExcludedItems,
recursive_search: bool,
delete_method: DeleteMethod,
stopped_search: bool,
2020-09-27 08:50:16 +13:00
}
impl EmptyFiles {
pub fn new() -> Self {
Self {
2020-09-27 08:50:16 +13:00
text_messages: Messages::new(),
information: Info::new(),
recursive_search: true,
allowed_extensions: Extensions::new(),
directories: Directories::new(),
excluded_items: ExcludedItems::new(),
empty_files: vec![],
delete_method: DeleteMethod::None,
stopped_search: false,
2020-09-27 08:50:16 +13:00
}
}
/// Finding empty files, save results to internal struct variables
pub fn find_empty_files(&mut self, stop_receiver: Option<&Receiver<()>>, progress_sender: Option<&futures::channel::mpsc::UnboundedSender<ProgressData>>) {
2020-09-27 08:50:16 +13:00
self.directories.optimize_directories(self.recursive_search, &mut self.text_messages);
2020-12-02 22:25:27 +13:00
if !self.check_files(stop_receiver, progress_sender) {
self.stopped_search = true;
return;
}
2020-09-27 08:50:16 +13:00
self.delete_files();
self.debug_print();
}
pub fn get_stopped_search(&self) -> bool {
self.stopped_search
}
pub const fn get_empty_files(&self) -> &Vec<FileEntry> {
&self.empty_files
}
pub const fn get_text_messages(&self) -> &Messages {
2020-09-27 08:50:16 +13:00
&self.text_messages
}
pub const fn get_information(&self) -> &Info {
2020-09-27 08:50:16 +13:00
&self.information
}
pub fn set_delete_method(&mut self, delete_method: DeleteMethod) {
self.delete_method = delete_method;
}
pub fn set_recursive_search(&mut self, recursive_search: bool) {
self.recursive_search = recursive_search;
}
pub fn set_included_directory(&mut self, included_directory: Vec<PathBuf>) -> bool {
2020-09-27 08:50:16 +13:00
self.directories.set_included_directory(included_directory, &mut self.text_messages)
}
pub fn set_excluded_directory(&mut self, excluded_directory: Vec<PathBuf>) {
2020-09-27 08:50:16 +13:00
self.directories.set_excluded_directory(excluded_directory, &mut self.text_messages);
}
pub fn set_allowed_extensions(&mut self, allowed_extensions: String) {
self.allowed_extensions.set_allowed_extensions(allowed_extensions, &mut self.text_messages);
}
pub fn set_excluded_items(&mut self, excluded_items: Vec<String>) {
2020-09-27 08:50:16 +13:00
self.excluded_items.set_excluded_items(excluded_items, &mut self.text_messages);
}
/// Check files for any with size == 0
fn check_files(&mut self, stop_receiver: Option<&Receiver<()>>, progress_sender: Option<&futures::channel::mpsc::UnboundedSender<ProgressData>>) -> bool {
2020-09-27 08:50:16 +13:00
let start_time: SystemTime = SystemTime::now();
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
2020-09-27 08:50:16 +13:00
// Add root folders for finding
for id in &self.directories.included_directories {
folders_to_check.push(id.clone());
2020-09-27 08:50:16 +13:00
}
2020-12-02 22:25:27 +13:00
//// PROGRESS THREAD START
const LOOP_DURATION: u32 = 200; //in ms
let progress_thread_run = Arc::new(AtomicBool::new(true));
let atomic_file_counter = Arc::new(AtomicUsize::new(0));
let progress_thread_handle = if let Some(progress_sender) = progress_sender {
let progress_send = progress_sender.clone();
2020-12-02 22:25:27 +13:00
let progress_thread_run = progress_thread_run.clone();
let atomic_file_counter = atomic_file_counter.clone();
thread::spawn(move || loop {
2020-12-02 22:25:27 +13:00
progress_send
.unbounded_send(ProgressData {
2020-12-02 22:25:27 +13:00
current_stage: 0,
max_stage: 0,
files_checked: atomic_file_counter.load(Ordering::Relaxed) as usize,
})
.unwrap();
if !progress_thread_run.load(Ordering::Relaxed) {
break;
}
sleep(Duration::from_millis(LOOP_DURATION as u64));
})
2020-12-02 22:25:27 +13:00
} else {
thread::spawn(|| {})
};
2020-12-02 22:25:27 +13:00
//// PROGRESS THREAD END
2020-09-27 08:50:16 +13:00
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() {
2020-12-02 22:25:27 +13:00
// End thread which send info to gui
progress_thread_run.store(false, Ordering::Relaxed);
progress_thread_handle.join().unwrap();
return false;
}
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![];
// Read current dir childrens
let read_dir = match fs::read_dir(&current_folder) {
Ok(t) => t,
Err(e) => {
warnings.push(format!("Cannot open dir {}, reason {}", current_folder.display(), e));
return (dir_result, warnings, fe_result);
}
};
// Check every sub folder/file/link etc.
'dir: for entry in read_dir {
let entry_data = match entry {
Ok(t) => t,
Err(e) => {
warnings.push(format!("Cannot read entry in dir {}, reason {}", current_folder.display(), e));
continue 'dir;
}
2020-09-27 08:50:16 +13:00
};
let metadata: Metadata = match entry_data.metadata() {
Ok(t) => t,
Err(e) => {
warnings.push(format!("Cannot read metadata in dir {}, reason {}", current_folder.display(), e));
continue 'dir;
}
};
if metadata.is_dir() {
if !self.recursive_search {
continue 'dir;
}
let next_folder = current_folder.join(entry_data.file_name());
if self.directories.is_excluded(&next_folder) {
continue 'dir;
}
if self.excluded_items.is_excluded(&next_folder) {
continue 'dir;
}
dir_result.push(next_folder);
} else if metadata.is_file() {
atomic_file_counter.fetch_add(1, Ordering::Relaxed);
let file_name_lowercase: String = match entry_data.file_name().into_string() {
Ok(t) => t,
Err(_inspected) => {
warnings.push(format!("File {:?} has not valid UTF-8 name", entry_data));
continue 'dir;
}
}
.to_lowercase();
if !self.allowed_extensions.matches_filename(&file_name_lowercase) {
continue 'dir;
}
if metadata.len() == 0 {
let current_file_name = current_folder.join(entry_data.file_name());
if self.excluded_items.is_excluded(&current_file_name) {
continue 'dir;
}
let fe: FileEntry = FileEntry {
path: current_file_name.clone(),
modified_date: match metadata.modified() {
Ok(t) => match t.duration_since(UNIX_EPOCH) {
Ok(d) => d.as_secs(),
Err(_inspected) => {
warnings.push(format!("File {} seems to be modified before Unix Epoch.", current_file_name.display()));
0
}
},
Err(e) => {
warnings.push(format!("Unable to get modification date from file {}, reason {}", current_file_name.display(), e));
0
}
},
};
fe_result.push(fe);
}
}
2020-09-27 08:50:16 +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);
self.text_messages.warnings.extend(warnings);
for fe in fe_result {
self.empty_files.push(fe);
2020-09-27 08:50:16 +13:00
}
}
}
self.information.number_of_empty_files = self.empty_files.len();
2020-12-02 22:25:27 +13:00
// End thread which send info to gui
progress_thread_run.store(false, Ordering::Relaxed);
progress_thread_handle.join().unwrap();
2020-09-27 08:50:16 +13:00
Common::print_time(start_time, SystemTime::now(), "check_files_size".to_string());
true
2020-09-27 08:50:16 +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 file_entry in &self.empty_files {
if fs::remove_file(file_entry.path.clone()).is_err() {
self.text_messages.warnings.push(file_entry.path.display().to_string());
2020-09-27 08:50:16 +13:00
}
}
}
DeleteMethod::None => {
//Just do nothing
}
}
Common::print_time(start_time, SystemTime::now(), "delete_files".to_string());
}
}
2021-11-28 08:57:10 +13:00
2020-09-27 08:50:16 +13:00
impl Default for EmptyFiles {
fn default() -> Self {
Self::new()
}
}
impl DebugPrint for EmptyFiles {
#[allow(dead_code)]
#[allow(unreachable_code)]
/// Debugging printing - only available on debug build
fn debug_print(&self) {
#[cfg(not(debug_assertions))]
{
return;
}
println!("---------------DEBUG PRINT---------------");
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 removed files - {}", self.information.number_of_removed_files);
println!("Number of failed to remove files - {}", self.information.number_of_failed_to_remove_files);
println!("### Other");
println!("Empty list size - {}", self.empty_files.len());
println!("Excluded items - {:?}", self.excluded_items.items);
println!("Included directories - {:?}", self.directories.included_directories);
println!("Excluded directories - {:?}", self.directories.excluded_directories);
println!("Recursive search - {}", self.recursive_search);
2020-09-27 08:50:16 +13:00
println!("Delete Method - {:?}", self.delete_method);
println!("-----------------------------------------");
}
}
2021-11-28 08:57:10 +13:00
2020-09-27 08:50:16 +13:00
impl SaveResults for EmptyFiles {
fn save_results_to_file(&mut self, file_name: &str) -> bool {
let start_time: SystemTime = SystemTime::now();
let file_name: String = match file_name {
"" => "results.txt".to_string(),
k => k.to_string(),
};
2021-01-09 23:52:43 +13:00
let file_handler = match File::create(&file_name) {
2020-09-27 08:50:16 +13:00
Ok(t) => t,
Err(e) => {
self.text_messages.errors.push(format!("Failed to create file {}, reason {}", file_name, e));
2020-09-27 08:50:16 +13:00
return false;
}
};
2021-01-09 23:52:43 +13:00
let mut writer = BufWriter::new(file_handler);
2020-09-27 08:50:16 +13:00
if let Err(e) = writeln!(
2021-01-09 23:52:43 +13:00
writer,
"Results of searching {:?} with excluded directories {:?} and excluded items {:?}",
self.directories.included_directories, self.directories.excluded_directories, self.excluded_items.items
) {
self.text_messages.errors.push(format!("Failed to save results to file {}, reason {}", file_name, e));
return false;
2020-09-27 08:50:16 +13:00
}
if !self.empty_files.is_empty() {
2021-01-09 23:52:43 +13:00
writeln!(writer, "Found {} empty files.", self.information.number_of_empty_files).unwrap();
2020-09-27 08:50:16 +13:00
for file_entry in self.empty_files.iter() {
2021-01-09 23:52:43 +13:00
writeln!(writer, "{}", file_entry.path.display()).unwrap();
2020-09-27 08:50:16 +13:00
}
} else {
2021-01-09 23:52:43 +13:00
write!(writer, "Not found any empty files.").unwrap();
2020-09-27 08:50:16 +13:00
}
Common::print_time(start_time, SystemTime::now(), "save_results_to_file".to_string());
true
}
}
2021-11-28 08:57:10 +13:00
2020-09-27 08:50:16 +13:00
impl PrintResults for EmptyFiles {
/// Print information's about duplicated entries
/// Only needed for CLI
fn print_results(&self) {
let start_time: SystemTime = SystemTime::now();
println!("Found {} empty files.\n", self.information.number_of_empty_files);
for file_entry in self.empty_files.iter() {
println!("{}", file_entry.path.display());
2020-09-27 08:50:16 +13:00
}
Common::print_time(start_time, SystemTime::now(), "print_entries".to_string());
}
}