2021-04-08 20:05:11 +05:30
|
|
|
/*
|
|
|
|
* Copyright (C) 2021 Aravinth Manivannan <realaravinth@batsense.net>
|
|
|
|
*
|
|
|
|
* Use of this source code is governed by the Apache 2.0 and/or the MIT
|
|
|
|
* License.
|
|
|
|
*/
|
|
|
|
|
|
|
|
use std::io::Error;
|
|
|
|
use std::path::Path;
|
|
|
|
use std::{fs, path::PathBuf};
|
|
|
|
|
|
|
|
use derive_builder::Builder;
|
|
|
|
use walkdir::WalkDir;
|
|
|
|
|
2021-04-08 20:45:38 +05:30
|
|
|
use crate::map::Files;
|
|
|
|
|
2021-04-08 20:05:11 +05:30
|
|
|
#[derive(Debug, Clone, Builder)]
|
|
|
|
pub struct Buster {
|
|
|
|
// source directory
|
|
|
|
#[builder(setter(into))]
|
|
|
|
source: String,
|
|
|
|
// mime_types for hashing
|
|
|
|
mime_types: Vec<mime::Mime>,
|
|
|
|
// directory for writing results
|
|
|
|
#[builder(setter(into))]
|
|
|
|
result: String,
|
|
|
|
// copy other non-hashed files from source dire to result dir?
|
|
|
|
copy: bool,
|
|
|
|
follow_links: bool,
|
|
|
|
}
|
|
|
|
|
|
|
|
impl Buster {
|
|
|
|
pub fn init(&self) -> Result<(), Error> {
|
|
|
|
let res = Path::new(&self.result);
|
|
|
|
if res.exists() {
|
|
|
|
fs::remove_dir_all(&self.result).unwrap();
|
|
|
|
}
|
|
|
|
|
|
|
|
fs::create_dir(&self.result).unwrap();
|
|
|
|
self.create_dir_structure(Path::new(&self.source))?;
|
|
|
|
Ok(())
|
|
|
|
}
|
|
|
|
|
2021-04-09 12:20:27 +05:30
|
|
|
fn hasher(payload: &[u8]) -> String {
|
2021-04-08 20:05:11 +05:30
|
|
|
use data_encoding::HEXUPPER;
|
|
|
|
use sha2::{Digest, Sha256};
|
|
|
|
let mut hasher = Sha256::new();
|
|
|
|
hasher.update(payload);
|
|
|
|
HEXUPPER.encode(&hasher.finalize())
|
|
|
|
}
|
|
|
|
|
|
|
|
// if mime types are common, then you shoud be fine using this
|
|
|
|
// use [hash] when when they aren't
|
|
|
|
//
|
|
|
|
// doesn't process files for which mime is not resolved
|
2021-04-08 20:45:38 +05:30
|
|
|
pub fn try_hash(&self) -> Result<Files, Error> {
|
|
|
|
let mut file_map: Files = Files::default();
|
2021-04-08 20:05:11 +05:30
|
|
|
for entry in WalkDir::new(&self.source)
|
|
|
|
.follow_links(self.follow_links)
|
|
|
|
.into_iter()
|
|
|
|
{
|
|
|
|
let entry = entry?;
|
|
|
|
let path = entry.path();
|
|
|
|
let path = Path::new(&path);
|
|
|
|
|
|
|
|
for mime_type in self.mime_types.iter() {
|
|
|
|
if let Some(file_mime) = mime_guess::from_path(path).first() {
|
|
|
|
if &file_mime == mime_type {
|
|
|
|
let contents = Self::read_to_string(&path).unwrap();
|
|
|
|
let hash = Self::hasher(&contents);
|
|
|
|
let new_name = format!(
|
|
|
|
"{}.{}.{}",
|
|
|
|
path.file_stem().unwrap().to_str().unwrap(),
|
|
|
|
hash,
|
|
|
|
path.extension().unwrap().to_str().unwrap()
|
|
|
|
);
|
|
|
|
self.copy(path, &new_name);
|
|
|
|
|
|
|
|
let (source, destination) = self.gen_map(path, &&new_name);
|
2021-04-08 22:00:41 +05:30
|
|
|
let _ = file_map.add(
|
|
|
|
source.to_str().unwrap().into(),
|
|
|
|
destination.to_str().unwrap().into(),
|
|
|
|
);
|
2021-04-08 20:05:11 +05:30
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
Ok(file_map)
|
|
|
|
}
|
|
|
|
|
|
|
|
// panics when mimetypes are detected. This way you'll know which files are ignored
|
|
|
|
// from processing
|
2021-04-08 20:45:38 +05:30
|
|
|
pub fn hash(&self) -> Result<Files, Error> {
|
|
|
|
let mut file_map: Files = Files::default();
|
2021-04-08 20:05:11 +05:30
|
|
|
|
|
|
|
for entry in WalkDir::new(&self.source)
|
|
|
|
.follow_links(self.follow_links)
|
|
|
|
.into_iter()
|
|
|
|
{
|
|
|
|
let entry = entry?;
|
|
|
|
|
|
|
|
let path = entry.path();
|
|
|
|
if !path.is_dir() {
|
|
|
|
let path = Path::new(&path);
|
|
|
|
|
|
|
|
for mime_type in self.mime_types.iter() {
|
|
|
|
let file_mime = mime_guess::from_path(path)
|
|
|
|
.first()
|
|
|
|
.expect(&format!("couldn't resolve MIME for file: {:?}", &path));
|
|
|
|
if &file_mime == mime_type {
|
|
|
|
let contents = Self::read_to_string(&path).unwrap();
|
|
|
|
let hash = Self::hasher(&contents);
|
|
|
|
let new_name = format!(
|
|
|
|
"{}.{}.{}",
|
|
|
|
path.file_stem().unwrap().to_str().unwrap(),
|
|
|
|
hash,
|
|
|
|
path.extension().unwrap().to_str().unwrap()
|
|
|
|
);
|
|
|
|
self.copy(path, &new_name);
|
|
|
|
let (source, destination) = self.gen_map(path, &&new_name);
|
2021-04-08 22:00:41 +05:30
|
|
|
let _ = file_map.add(
|
|
|
|
source.to_str().unwrap().into(),
|
|
|
|
destination.to_str().unwrap().into(),
|
|
|
|
);
|
2021-04-08 20:05:11 +05:30
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
Ok(file_map)
|
|
|
|
}
|
|
|
|
|
2021-04-09 12:20:27 +05:30
|
|
|
fn read_to_string(path: &Path) -> Result<Vec<u8>, Error> {
|
2021-04-08 20:05:11 +05:30
|
|
|
use std::fs::File;
|
2021-04-09 12:20:27 +05:30
|
|
|
use std::io::Read;
|
2021-04-08 20:05:11 +05:30
|
|
|
|
2021-04-09 12:20:27 +05:30
|
|
|
let mut file_content = Vec::new();
|
|
|
|
let mut file = File::open(path)?;
|
|
|
|
file.read_to_end(&mut file_content).expect("Unable to read");
|
|
|
|
Ok(file_content)
|
2021-04-08 20:05:11 +05:30
|
|
|
}
|
|
|
|
|
|
|
|
fn gen_map<'a>(&self, source: &'a Path, name: &str) -> (&'a Path, PathBuf) {
|
|
|
|
let rel_location = source.strip_prefix(&self.source).unwrap().parent().unwrap();
|
|
|
|
let destination = Path::new(&self.result).join(rel_location).join(name);
|
|
|
|
(source, destination)
|
|
|
|
}
|
|
|
|
|
|
|
|
fn copy(&self, source: &Path, name: &str) {
|
|
|
|
let rel_location = source.strip_prefix(&self.source).unwrap().parent().unwrap();
|
|
|
|
let destination = Path::new(&self.result).join(rel_location).join(name);
|
|
|
|
fs::copy(source, &destination).unwrap();
|
|
|
|
}
|
|
|
|
|
|
|
|
fn create_dir_structure(&self, path: &Path) -> Result<(), Error> {
|
|
|
|
for entry in WalkDir::new(&path)
|
|
|
|
.follow_links(self.follow_links)
|
|
|
|
.into_iter()
|
|
|
|
{
|
|
|
|
let entry = entry?;
|
|
|
|
let entry_path = entry.path();
|
|
|
|
let entry_path = Path::new(&entry_path);
|
|
|
|
|
|
|
|
if entry_path.is_dir() && path != entry_path {
|
|
|
|
Self::create_dir_structure(&self, entry_path)?;
|
|
|
|
} else {
|
|
|
|
if entry_path.is_dir() {
|
|
|
|
let rel_location = entry_path.strip_prefix(&self.source).unwrap();
|
|
|
|
let destination = Path::new(&self.result).join(rel_location);
|
|
|
|
if !destination.exists() {
|
|
|
|
fs::create_dir(destination)?
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
Ok(())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#[cfg(test)]
|
2021-04-08 22:00:41 +05:30
|
|
|
pub mod tests {
|
2021-04-08 20:05:11 +05:30
|
|
|
use super::*;
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn hasher_works() {
|
|
|
|
let types = vec![
|
|
|
|
mime::IMAGE_PNG,
|
|
|
|
mime::IMAGE_SVG,
|
|
|
|
mime::IMAGE_JPEG,
|
|
|
|
mime::IMAGE_GIF,
|
|
|
|
];
|
|
|
|
|
|
|
|
let config = BusterBuilder::default()
|
|
|
|
.source("./dist")
|
|
|
|
.result("./prod")
|
|
|
|
.mime_types(types)
|
|
|
|
.copy(true)
|
|
|
|
.follow_links(true)
|
|
|
|
.build()
|
|
|
|
.unwrap();
|
|
|
|
|
|
|
|
config.init().unwrap();
|
2021-04-08 20:45:38 +05:30
|
|
|
let mut files = config.hash().unwrap();
|
2021-04-08 20:05:11 +05:30
|
|
|
|
2021-04-08 20:45:38 +05:30
|
|
|
for (k, v) in files.map.drain() {
|
2021-04-08 20:05:11 +05:30
|
|
|
let src = Path::new(&k);
|
|
|
|
let dest = Path::new(&v);
|
|
|
|
|
|
|
|
assert_eq!(src.exists(), dest.exists());
|
|
|
|
}
|
2021-04-08 22:00:41 +05:30
|
|
|
cleanup(&config);
|
2021-04-08 20:05:11 +05:30
|
|
|
}
|
2021-04-08 20:14:53 +05:30
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn try_hash_works() {
|
|
|
|
let types = vec![
|
|
|
|
mime::IMAGE_PNG,
|
|
|
|
mime::IMAGE_SVG,
|
|
|
|
mime::IMAGE_JPEG,
|
|
|
|
mime::IMAGE_GIF,
|
|
|
|
];
|
|
|
|
|
|
|
|
let config = BusterBuilder::default()
|
|
|
|
.source("./dist")
|
|
|
|
.result("/tmp/prod")
|
|
|
|
.mime_types(types)
|
|
|
|
.copy(true)
|
|
|
|
.follow_links(true)
|
|
|
|
.build()
|
|
|
|
.unwrap();
|
|
|
|
|
|
|
|
config.init().unwrap();
|
2021-04-08 22:11:26 +05:30
|
|
|
let mut files = config.try_hash().unwrap();
|
2021-04-08 20:14:53 +05:30
|
|
|
|
2021-04-08 20:45:38 +05:30
|
|
|
for (k, v) in files.map.drain() {
|
2021-04-08 20:14:53 +05:30
|
|
|
let src = Path::new(&k);
|
|
|
|
let dest = Path::new(&v);
|
|
|
|
|
|
|
|
assert_eq!(src.exists(), dest.exists());
|
|
|
|
}
|
2021-04-08 22:00:41 +05:30
|
|
|
|
|
|
|
cleanup(&config);
|
|
|
|
}
|
|
|
|
|
|
|
|
pub fn cleanup(config: &Buster) {
|
|
|
|
let _ = fs::remove_dir_all(&config.result);
|
2021-04-08 20:14:53 +05:30
|
|
|
}
|
2021-04-08 20:05:11 +05:30
|
|
|
}
|