Add support for finding similar images to CLI (#66)
This commit is contained in:
Rafał Mikrut
GitHub
parent
acfecd7ca3
commit
3dd203f246
96
Cargo.lock
generated
96
Cargo.lock
generated
|
|
@ -97,6 +97,12 @@ version = "0.2.6"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "1b20b618342cf9891c292c4f5ac2cde7287cc5c87e87e9c769d617793607dec1"
|
||||
|
||||
[[package]]
|
||||
name = "base64"
|
||||
version = "0.11.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "b41b7ea54a0c9d92199de89e20e58d49f02f8e699814ef3fdf266f6f748d15c7"
|
||||
|
||||
[[package]]
|
||||
name = "base64"
|
||||
version = "0.12.3"
|
||||
|
|
@ -109,6 +115,12 @@ version = "1.2.1"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "cf1de2fe8c75bc145a2f577add951f8134889b4795d47466a54a5c846d691693"
|
||||
|
||||
[[package]]
|
||||
name = "bk-tree"
|
||||
version = "0.3.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "5488039ea2c6de8668351415e39a0218a8955bffadcff0cf01d1293a20854584"
|
||||
|
||||
[[package]]
|
||||
name = "blake2b_simd"
|
||||
version = "0.5.10"
|
||||
|
|
@ -343,9 +355,12 @@ dependencies = [
|
|||
name = "czkawka_core"
|
||||
version = "1.1.0"
|
||||
dependencies = [
|
||||
"bk-tree",
|
||||
"blake3",
|
||||
"crossbeam-channel",
|
||||
"humansize",
|
||||
"image",
|
||||
"img_hash",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
|
|
@ -360,6 +375,7 @@ dependencies = [
|
|||
"glib",
|
||||
"gtk",
|
||||
"humansize",
|
||||
"open",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
|
|
@ -857,6 +873,19 @@ dependencies = [
|
|||
"tiff",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "img_hash"
|
||||
version = "3.1.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "8df6c5bc88f37a165c63143e38924f691246fc77f12de0cbd126fe0c8ca3527b"
|
||||
dependencies = [
|
||||
"base64 0.11.0",
|
||||
"image",
|
||||
"rustdct",
|
||||
"serde",
|
||||
"transpose 0.2.0",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "inflate"
|
||||
version = "0.4.5"
|
||||
|
|
@ -1013,6 +1042,16 @@ dependencies = [
|
|||
"num-traits",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "num-complex"
|
||||
version = "0.2.4"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "b6b19411a9719e753aff12e5187b74d60d3dc449ec3f4dc21e3989c3f554bc95"
|
||||
dependencies = [
|
||||
"autocfg 1.0.1",
|
||||
"num-traits",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "num-integer"
|
||||
version = "0.1.43"
|
||||
|
|
@ -1070,6 +1109,15 @@ version = "1.4.1"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "260e51e7efe62b592207e9e13a68e43692a7a279171d6ba57abd208bf23645ad"
|
||||
|
||||
[[package]]
|
||||
name = "open"
|
||||
version = "1.4.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "7c283bf0114efea9e42f1a60edea9859e8c47528eae09d01df4b29c1e489cc48"
|
||||
dependencies = [
|
||||
"winapi",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "orbclient"
|
||||
version = "0.3.27"
|
||||
|
|
@ -1636,7 +1684,7 @@ version = "0.6.2"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "f8a58080b7bb83b2ea28c3b7a9a994fd5e310330b7c8ca5258d99b98128ecfe4"
|
||||
dependencies = [
|
||||
"base64",
|
||||
"base64 0.12.3",
|
||||
"bitflags",
|
||||
"serde",
|
||||
]
|
||||
|
|
@ -1647,7 +1695,7 @@ version = "0.8.2"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "9dab61250775933275e84053ac235621dfb739556d5c54a2f2e9313b7cf43a19"
|
||||
dependencies = [
|
||||
"base64",
|
||||
"base64 0.12.3",
|
||||
"blake2b_simd",
|
||||
"constant_time_eq",
|
||||
"crossbeam-utils",
|
||||
|
|
@ -1672,6 +1720,28 @@ dependencies = [
|
|||
"semver",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "rustdct"
|
||||
version = "0.4.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "ef4d167674b4cf68c2114bdbcd34c95aa9071652b73b0f43b19298f1d2780b7d"
|
||||
dependencies = [
|
||||
"rustfft",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "rustfft"
|
||||
version = "3.0.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "77008ed59a8923c8b4ac2e5eaa6d28fbe893d3b9515098a4a5fc7767d6430fe5"
|
||||
dependencies = [
|
||||
"num-complex",
|
||||
"num-integer",
|
||||
"num-traits",
|
||||
"strength_reduce",
|
||||
"transpose 0.1.0",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "rusttype"
|
||||
version = "0.8.3"
|
||||
|
|
@ -1865,6 +1935,12 @@ version = "0.1.5"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "213701ba3370744dcd1a12960caa4843b3d68b4d1c0a5d575e0d65b2ee9d16c0"
|
||||
|
||||
[[package]]
|
||||
name = "strength_reduce"
|
||||
version = "0.2.3"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "a3ff2f71c82567c565ba4b3009a9350a96a7269eaa4001ebedae926230bc2254"
|
||||
|
||||
[[package]]
|
||||
name = "strsim"
|
||||
version = "0.8.0"
|
||||
|
|
@ -2036,6 +2112,22 @@ dependencies = [
|
|||
"serde",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "transpose"
|
||||
version = "0.1.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "643e21580bb0627c7bb09e5cedbb42c8705b19d012de593ed6b0309270b3cd1e"
|
||||
|
||||
[[package]]
|
||||
name = "transpose"
|
||||
version = "0.2.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "c3311ef71dea6a1fd6bf5bfc10ec5b4bef6174048f6b481dbc6ce915ff48c0a0"
|
||||
dependencies = [
|
||||
"num-integer",
|
||||
"strength_reduce",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "typed-arena"
|
||||
version = "2.0.1"
|
||||
|
|
|
|||
|
|
@ -13,7 +13,7 @@ pub enum Commands {
|
|||
#[structopt(flatten)]
|
||||
excluded_items: ExcludedItems,
|
||||
#[structopt(short, long, parse(try_from_str = parse_minimal_file_size), default_value = "1024", help = "Minimum size in bytes", long_help = "Minimum size of checked files in bytes, assigning bigger value may speed up searching")]
|
||||
min_size: u64,
|
||||
minimal_file_size: u64,
|
||||
#[structopt(flatten)]
|
||||
allowed_extensions: AllowedExtensions,
|
||||
#[structopt(short, long, default_value = "HASH", parse(try_from_str = parse_checking_method), help = "Search method (SIZE, HASH, HASHMB)", long_help = "Methods to search files.\nSIZE - The fastest method, checking by the file's size,\nHASHMB - More accurate but slower, checking by the hash of the file's first mibibyte or\nHASH - The slowest method, checking by the hash of the entire file")]
|
||||
|
|
@ -83,6 +83,21 @@ pub enum Commands {
|
|||
#[structopt(flatten)]
|
||||
not_recursive: NotRecursive,
|
||||
},
|
||||
#[structopt(name = "ima", about = "Finds similar images", help_message = HELP_MESSAGE, after_help = "EXAMPLE:\n czkawka ima -d /home/rafal/ -E */.git */tmp* *Pulpit -f results.txt")]
|
||||
SimilarImages {
|
||||
#[structopt(flatten)]
|
||||
directories: Directories,
|
||||
#[structopt(flatten)]
|
||||
excluded_directories: ExcludedDirectories,
|
||||
#[structopt(short, long, parse(try_from_str = parse_minimal_file_size), default_value = "16384", help = "Minimum size in bytes", long_help = "Minimum size of checked files in bytes, assigning bigger value may speed up searching")]
|
||||
minimal_file_size: u64,
|
||||
#[structopt(flatten)]
|
||||
excluded_items: ExcludedItems,
|
||||
#[structopt(flatten)]
|
||||
file_to_save: FileToSave,
|
||||
#[structopt(flatten)]
|
||||
not_recursive: NotRecursive,
|
||||
},
|
||||
}
|
||||
|
||||
#[derive(Debug, StructOpt)]
|
||||
|
|
@ -158,9 +173,9 @@ fn parse_delete_method(src: &str) -> Result<DeleteMethod, &'static str> {
|
|||
|
||||
fn parse_minimal_file_size(src: &str) -> Result<u64, String> {
|
||||
match src.parse::<u64>() {
|
||||
Ok(min_size) => {
|
||||
if min_size > 0 {
|
||||
Ok(min_size)
|
||||
Ok(minimal_file_size) => {
|
||||
if minimal_file_size > 0 {
|
||||
Ok(minimal_file_size)
|
||||
} else {
|
||||
Err("Minimum file size must be at least 1 byte".to_string())
|
||||
}
|
||||
|
|
|
|||
|
|
@ -10,6 +10,7 @@ use czkawka_core::{
|
|||
duplicate::DuplicateFinder,
|
||||
empty_files::{self, EmptyFiles},
|
||||
empty_folder::EmptyFolder,
|
||||
similar_files::SimilarImages,
|
||||
temporary::{self, Temporary},
|
||||
};
|
||||
use std::{path::PathBuf, process};
|
||||
|
|
@ -31,7 +32,7 @@ fn main() {
|
|||
directories,
|
||||
excluded_directories,
|
||||
excluded_items,
|
||||
min_size,
|
||||
minimal_file_size,
|
||||
allowed_extensions,
|
||||
search_method,
|
||||
delete_method,
|
||||
|
|
@ -43,7 +44,7 @@ fn main() {
|
|||
df.set_included_directory(path_list_to_str(directories.directories));
|
||||
df.set_excluded_directory(path_list_to_str(excluded_directories.excluded_directories));
|
||||
df.set_excluded_items(path_list_to_str(excluded_items.excluded_items));
|
||||
df.set_minimal_file_size(min_size);
|
||||
df.set_minimal_file_size(minimal_file_size);
|
||||
df.set_allowed_extensions(allowed_extensions.allowed_extensions.join(","));
|
||||
df.set_check_method(search_method);
|
||||
df.set_delete_method(delete_method);
|
||||
|
|
@ -178,5 +179,34 @@ fn main() {
|
|||
tf.print_results();
|
||||
tf.get_text_messages().print_messages();
|
||||
}
|
||||
Commands::SimilarImages {
|
||||
directories,
|
||||
excluded_directories,
|
||||
excluded_items,
|
||||
file_to_save,
|
||||
minimal_file_size,
|
||||
not_recursive,
|
||||
} => {
|
||||
let mut sf = SimilarImages::new();
|
||||
|
||||
sf.set_included_directory(path_list_to_str(directories.directories));
|
||||
sf.set_excluded_directory(path_list_to_str(excluded_directories.excluded_directories));
|
||||
sf.set_excluded_items(path_list_to_str(excluded_items.excluded_items));
|
||||
sf.set_minimal_file_size(minimal_file_size);
|
||||
sf.set_recursive_search(!not_recursive.not_recursive);
|
||||
|
||||
sf.find_similar_images(None);
|
||||
|
||||
if let Some(file_name) = file_to_save.file_name() {
|
||||
if !sf.save_results_to_file(file_name) {
|
||||
sf.get_text_messages().print_messages();
|
||||
process::exit(1);
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(not(debug_assertions))] // This will show too much probably unnecessary data to debug, comment line only if needed
|
||||
sf.print_results();
|
||||
sf.get_text_messages().print_messages();
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -13,4 +13,10 @@ repository = "https://github.com/qarmin/czkawka"
|
|||
humansize = "1"
|
||||
blake3 = "0.3"
|
||||
#rayon = "1"
|
||||
crossbeam-channel = "0.4.4"
|
||||
crossbeam-channel = "0.4.4"
|
||||
|
||||
|
||||
# Needed by similar images
|
||||
img_hash = "3.1"
|
||||
bk-tree = "0.3"
|
||||
image = "0.23"
|
||||
|
|
@ -325,7 +325,7 @@ impl SaveResults for BigFile {
|
|||
}
|
||||
}
|
||||
} else {
|
||||
write!(file, "Not found any empty folders.").unwrap();
|
||||
write!(file, "Not found any files.").unwrap();
|
||||
}
|
||||
Common::print_time(start_time, SystemTime::now(), "save_results_to_file".to_string());
|
||||
true
|
||||
|
|
|
|||
|
|
@ -10,5 +10,6 @@ pub mod common_extensions;
|
|||
pub mod common_items;
|
||||
pub mod common_messages;
|
||||
pub mod common_traits;
|
||||
pub mod similar_files;
|
||||
|
||||
pub const CZKAWKA_VERSION: &str = env!("CARGO_PKG_VERSION");
|
||||
|
|
|
|||
421
czkawka_core/src/similar_files.rs
Normal file
421
czkawka_core/src/similar_files.rs
Normal file
|
|
@ -0,0 +1,421 @@
|
|||
use crate::common::Common;
|
||||
use crate::common_directory::Directories;
|
||||
use crate::common_items::ExcludedItems;
|
||||
use crate::common_messages::Messages;
|
||||
use crate::common_traits::{DebugPrint, PrintResults, SaveResults};
|
||||
use bk_tree::{metrics, BKTree};
|
||||
use crossbeam_channel::Receiver;
|
||||
use humansize::{file_size_opts as options, FileSize};
|
||||
use img_hash::HasherConfig;
|
||||
use std::collections::HashMap;
|
||||
use std::fs;
|
||||
use std::fs::{File, Metadata};
|
||||
use std::io::Write;
|
||||
use std::path::PathBuf;
|
||||
use std::time::{SystemTime, UNIX_EPOCH};
|
||||
|
||||
#[derive(Clone, Eq, PartialEq)]
|
||||
pub enum Similarity {
|
||||
None,
|
||||
Small,
|
||||
Medium,
|
||||
High,
|
||||
VeryHigh,
|
||||
}
|
||||
|
||||
#[derive(Clone)]
|
||||
pub struct FileEntry {
|
||||
pub path: PathBuf,
|
||||
pub size: u64,
|
||||
pub modified_date: u64,
|
||||
pub similarity: Similarity,
|
||||
}
|
||||
#[derive(Clone)]
|
||||
pub struct StructSimilar {
|
||||
pub base_image: FileEntry,
|
||||
pub similar_images: Vec<FileEntry>,
|
||||
}
|
||||
|
||||
/// Struct to store most basics info about all folder
|
||||
pub struct SimilarImages {
|
||||
information: Info,
|
||||
text_messages: Messages,
|
||||
directories: Directories,
|
||||
excluded_items: ExcludedItems,
|
||||
bktree: BKTree<String>,
|
||||
similar_vectors: Vec<StructSimilar>,
|
||||
recursive_search: bool,
|
||||
minimal_file_size: u64,
|
||||
image_hashes: HashMap<String, Vec<FileEntry>>, // Hashmap with image hashes and Vector with names of files
|
||||
stopped_search: bool,
|
||||
}
|
||||
|
||||
/// Info struck with helpful information's about results
|
||||
#[derive(Default)]
|
||||
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 size_of_checked_images: u64,
|
||||
pub lost_space: u64,
|
||||
pub number_of_removed_files: usize,
|
||||
pub number_of_failed_to_remove_files: usize,
|
||||
pub gained_space: u64,
|
||||
}
|
||||
impl Info {
|
||||
pub fn new() -> Self {
|
||||
Default::default()
|
||||
}
|
||||
}
|
||||
|
||||
/// Method implementation for EmptyFolder
|
||||
impl SimilarImages {
|
||||
/// New function providing basics values
|
||||
pub fn new() -> Self {
|
||||
Self {
|
||||
information: Default::default(),
|
||||
text_messages: Messages::new(),
|
||||
directories: Directories::new(),
|
||||
excluded_items: Default::default(),
|
||||
bktree: BKTree::new(metrics::Levenshtein),
|
||||
similar_vectors: vec![],
|
||||
recursive_search: true,
|
||||
minimal_file_size: 1024 * 16, // 16 KB should be enough to exclude too small images from search
|
||||
image_hashes: Default::default(),
|
||||
stopped_search: false,
|
||||
}
|
||||
}
|
||||
|
||||
pub fn get_stopped_search(&self) -> bool {
|
||||
self.stopped_search
|
||||
}
|
||||
|
||||
pub const fn get_text_messages(&self) -> &Messages {
|
||||
&self.text_messages
|
||||
}
|
||||
|
||||
pub const fn get_information(&self) -> &Info {
|
||||
&self.information
|
||||
}
|
||||
|
||||
pub fn set_recursive_search(&mut self, recursive_search: bool) {
|
||||
self.recursive_search = recursive_search;
|
||||
}
|
||||
|
||||
pub fn set_minimal_file_size(&mut self, minimal_file_size: u64) {
|
||||
self.minimal_file_size = match minimal_file_size {
|
||||
0 => 1,
|
||||
t => t,
|
||||
};
|
||||
}
|
||||
|
||||
/// Public function used by CLI to search for empty folders
|
||||
pub fn find_similar_images(&mut self, rx: Option<&Receiver<()>>) {
|
||||
self.directories.optimize_directories(true, &mut self.text_messages);
|
||||
if !self.check_for_similar_images(rx) {
|
||||
self.stopped_search = true;
|
||||
return;
|
||||
}
|
||||
// if self.delete_folders {
|
||||
// self.delete_empty_folders();
|
||||
// }
|
||||
self.debug_print();
|
||||
}
|
||||
|
||||
// pub fn set_delete_folder(&mut self, delete_folder: bool) {
|
||||
// self.delete_folders = delete_folder;
|
||||
// }
|
||||
|
||||
/// Function to check if folder are empty.
|
||||
/// Parameter initial_checking for second check before deleting to be sure that checked folder is still empty
|
||||
fn check_for_similar_images(&mut self, rx: Option<&Receiver<()>>) -> bool {
|
||||
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
|
||||
|
||||
// Add root folders for finding
|
||||
for id in &self.directories.included_directories {
|
||||
folders_to_check.push(id.clone());
|
||||
}
|
||||
self.information.number_of_checked_folders += folders_to_check.len();
|
||||
|
||||
while !folders_to_check.is_empty() {
|
||||
if rx.is_some() && rx.unwrap().try_recv().is_ok() {
|
||||
return false;
|
||||
}
|
||||
let current_folder = folders_to_check.pop().unwrap();
|
||||
|
||||
// Read current dir, if permission are denied just go to next
|
||||
let read_dir = match fs::read_dir(¤t_folder) {
|
||||
Ok(t) => t,
|
||||
Err(_) => {
|
||||
self.text_messages.warnings.push(format!("Cannot open dir {}", current_folder.display()));
|
||||
continue;
|
||||
} // Permissions denied
|
||||
};
|
||||
|
||||
// Check every sub folder/file/link etc.
|
||||
'dir: for entry in read_dir {
|
||||
let entry_data = match entry {
|
||||
Ok(t) => t,
|
||||
Err(_) => {
|
||||
self.text_messages.warnings.push(format!("Cannot read entry in dir {}", current_folder.display()));
|
||||
continue;
|
||||
} //Permissions denied
|
||||
};
|
||||
let metadata: Metadata = match entry_data.metadata() {
|
||||
Ok(t) => t,
|
||||
Err(_) => {
|
||||
self.text_messages.warnings.push(format!("Cannot read metadata in dir {}", current_folder.display()));
|
||||
continue;
|
||||
} //Permissions denied
|
||||
};
|
||||
if metadata.is_dir() {
|
||||
self.information.number_of_checked_folders += 1;
|
||||
|
||||
if !self.recursive_search {
|
||||
continue;
|
||||
}
|
||||
|
||||
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;
|
||||
}
|
||||
|
||||
folders_to_check.push(next_folder);
|
||||
} else if metadata.is_file() {
|
||||
// let mut have_valid_extension: bool;
|
||||
let file_name_lowercase: String = match entry_data.file_name().into_string() {
|
||||
Ok(t) => t,
|
||||
Err(_) => continue,
|
||||
}
|
||||
.to_lowercase();
|
||||
|
||||
// Checking allowed image extensions
|
||||
let allowed_image_extensions = ["jpg", "png", "bmp"];
|
||||
if !allowed_image_extensions.iter().any(|e| file_name_lowercase.ends_with(e)) {
|
||||
self.information.number_of_ignored_files += 1;
|
||||
continue 'dir;
|
||||
}
|
||||
|
||||
// Checking files
|
||||
if metadata.len() >= self.minimal_file_size {
|
||||
let current_file_name = current_folder.join(entry_data.file_name());
|
||||
if self.excluded_items.is_excluded(¤t_file_name) {
|
||||
continue 'dir;
|
||||
}
|
||||
|
||||
// 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(),
|
||||
Err(_) => {
|
||||
self.text_messages.warnings.push(format!("File {} seems to be modified before Unix Epoch.", current_file_name.display()));
|
||||
0
|
||||
}
|
||||
},
|
||||
Err(_) => {
|
||||
self.text_messages.warnings.push(format!("Unable to get modification date from file {}", current_file_name.display()));
|
||||
continue;
|
||||
} // Permissions Denied
|
||||
},
|
||||
|
||||
similarity: Similarity::None,
|
||||
};
|
||||
let hasher = HasherConfig::new().to_hasher();
|
||||
let image = match image::open(current_file_name) {
|
||||
Ok(t) => t,
|
||||
Err(_) => continue 'dir, // Something is wrong with image
|
||||
};
|
||||
|
||||
let hash = hasher.hash_image(&image);
|
||||
let string_hash = hash.to_base64();
|
||||
|
||||
self.bktree.add(string_hash.clone());
|
||||
self.image_hashes.entry(string_hash.clone()).or_insert_with(Vec::<FileEntry>::new);
|
||||
self.image_hashes.get_mut(&string_hash).unwrap().push(fe);
|
||||
|
||||
self.information.size_of_checked_images += metadata.len();
|
||||
self.information.number_of_checked_files += 1;
|
||||
} else {
|
||||
// Probably this is symbolic links so we are free to ignore this
|
||||
self.information.number_of_ignored_files += 1;
|
||||
}
|
||||
} else {
|
||||
// Probably this is symbolic links so we are free to ignore this
|
||||
self.information.number_of_ignored_things += 1;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
let hash_map_modification = SystemTime::now();
|
||||
|
||||
let mut new_vector: Vec<StructSimilar> = Vec::new();
|
||||
for (string_hash, vec_file_entry) in &self.image_hashes {
|
||||
let vector_with_found_similar_hashes = self.bktree.find(string_hash.as_str(), 3).collect::<Vec<_>>();
|
||||
if vector_with_found_similar_hashes.len() == 1 && vec_file_entry.len() == 1 {
|
||||
// Exists only 1 unique picture, so there is no need to use it
|
||||
continue;
|
||||
}
|
||||
|
||||
let mut vec_similarity_struct: Vec<StructSimilar> = Vec::new();
|
||||
|
||||
for file_entry in vec_file_entry.iter() {
|
||||
let similar_struct = StructSimilar {
|
||||
base_image: file_entry.clone(),
|
||||
similar_images: vec_file_entry
|
||||
.iter()
|
||||
.filter(|x| x.path != file_entry.path)
|
||||
.map(|x| {
|
||||
let mut y = x.clone();
|
||||
y.similarity = Similarity::VeryHigh;
|
||||
y
|
||||
})
|
||||
.collect::<Vec<FileEntry>>(),
|
||||
};
|
||||
vec_similarity_struct.push(similar_struct);
|
||||
}
|
||||
|
||||
for (similarity, hash) in vector_with_found_similar_hashes.iter() {
|
||||
if *similarity == 0 && string_hash == *hash {
|
||||
// This was already readed before
|
||||
continue;
|
||||
} else if string_hash == *hash {
|
||||
panic!("I'm not sure if same hash can have distance > 0");
|
||||
}
|
||||
|
||||
for file_entry in self.image_hashes.get(*hash).unwrap() {
|
||||
let mut file_entry = file_entry.clone();
|
||||
file_entry.similarity = match similarity {
|
||||
0 => Similarity::VeryHigh,
|
||||
1 => Similarity::High,
|
||||
2 => Similarity::Medium,
|
||||
3 => Similarity::Small,
|
||||
_ => panic!("0-3 similarity levels are allowed, check if not added more."),
|
||||
};
|
||||
for similarity_struct in vec_similarity_struct.iter_mut() {
|
||||
similarity_struct.similar_images.push(file_entry.clone());
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
new_vector.append(&mut vec_similarity_struct);
|
||||
}
|
||||
|
||||
self.similar_vectors = new_vector;
|
||||
|
||||
#[allow(clippy::blocks_in_if_conditions)]
|
||||
Common::print_time(hash_map_modification, SystemTime::now(), "hash_map_modification(internal)".to_string());
|
||||
Common::print_time(start_time, SystemTime::now(), "check_for_similar_images".to_string());
|
||||
true
|
||||
}
|
||||
|
||||
/// Set included dir which needs to be relative, exists etc.
|
||||
pub fn set_included_directory(&mut self, included_directory: String) {
|
||||
self.directories.set_included_directory(included_directory, &mut self.text_messages);
|
||||
}
|
||||
|
||||
pub fn set_excluded_directory(&mut self, excluded_directory: String) {
|
||||
self.directories.set_excluded_directory(excluded_directory, &mut self.text_messages);
|
||||
}
|
||||
|
||||
pub fn set_excluded_items(&mut self, excluded_items: String) {
|
||||
self.excluded_items.set_excluded_items(excluded_items, &mut self.text_messages);
|
||||
}
|
||||
}
|
||||
impl Default for SimilarImages {
|
||||
fn default() -> Self {
|
||||
Self::new()
|
||||
}
|
||||
}
|
||||
|
||||
impl DebugPrint for SimilarImages {
|
||||
#[allow(dead_code)]
|
||||
#[allow(unreachable_code)]
|
||||
fn debug_print(&self) {
|
||||
#[cfg(not(debug_assertions))]
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
println!("---------------DEBUG PRINT---------------");
|
||||
println!("Number of all checked folders - {}", self.information.number_of_checked_folders);
|
||||
println!("Included directories - {:?}", self.directories.included_directories);
|
||||
println!("Checked images {} / Different photos {}", self.information.number_of_checked_files, self.image_hashes.len());
|
||||
println!(
|
||||
"Size of checked images {} ({} Bytes)",
|
||||
self.information.size_of_checked_images.file_size(options::BINARY).unwrap(),
|
||||
self.information.size_of_checked_images
|
||||
);
|
||||
println!("-----------------------------------------");
|
||||
}
|
||||
}
|
||||
impl SaveResults for SimilarImages {
|
||||
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(),
|
||||
};
|
||||
|
||||
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;
|
||||
}
|
||||
};
|
||||
|
||||
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;
|
||||
}
|
||||
|
||||
if !self.similar_vectors.is_empty() {
|
||||
write!(file, "{} images which have similar friends\n\n", self.similar_vectors.len()).unwrap();
|
||||
|
||||
for struct_similar in self.similar_vectors.iter() {
|
||||
writeln!(file, "Image {:?} have {} similar images", struct_similar.base_image.path, struct_similar.similar_images.len()).unwrap();
|
||||
for similar_picture in struct_similar.similar_images.iter() {
|
||||
writeln!(file, "{:?} - Similarity Level: {}", similar_picture.path, get_string_from_similarity(&similar_picture.similarity)).unwrap();
|
||||
}
|
||||
writeln!(file).unwrap();
|
||||
}
|
||||
} else {
|
||||
write!(file, "Not found any similar images.").unwrap();
|
||||
}
|
||||
Common::print_time(start_time, SystemTime::now(), "save_results_to_file".to_string());
|
||||
true
|
||||
}
|
||||
}
|
||||
impl PrintResults for SimilarImages {
|
||||
/// Prints basic info about empty folders // TODO print better
|
||||
fn print_results(&self) {
|
||||
if !self.similar_vectors.is_empty() {
|
||||
println!("Found {} images which have similar friends", self.similar_vectors.len());
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn get_string_from_similarity(similarity: &Similarity) -> &str {
|
||||
match similarity {
|
||||
Similarity::Small => "Small",
|
||||
Similarity::Medium => "Medium",
|
||||
Similarity::High => "High",
|
||||
Similarity::VeryHigh => "Very High",
|
||||
Similarity::None => panic!(),
|
||||
}
|
||||
}
|
||||
|
|
@ -19,6 +19,9 @@ chrono = "0.4"
|
|||
|
||||
crossbeam-channel = "0.4.4"
|
||||
|
||||
# For opening files
|
||||
open = "1.4.0"
|
||||
|
||||
[dependencies.gtk]
|
||||
version = "0.9.2"
|
||||
default-features = false # just in case
|
||||
|
|
|
|||
Loading…
Reference in a new issue