feat: performance improvements, fix zip

This commit is contained in:
DecDuck
2025-08-13 11:35:50 +10:00
parent 1665033fd9
commit dc3a420986
13 changed files with 504 additions and 227 deletions
+10 -8
View File
@@ -13,8 +13,7 @@ use napi::{
use serde_json::json;
use uuid::Uuid;
use crate::version::utils::create_backend_for_path;
use crate::version::{types::VersionBackend, utils::DropletHandler};
const CHUNK_SIZE: usize = 1024 * 1024 * 64;
@@ -36,15 +35,18 @@ pub fn call_alt_thread_func(tsfn: Arc<ThreadsafeFunction<()>>) -> Result<(), Str
}
#[napi]
pub fn generate_manifest(
pub fn generate_manifest<'a>(
droplet_handler: &mut DropletHandler,
dir: String,
progress_sfn: ThreadsafeFunction<i32>,
log_sfn: ThreadsafeFunction<String>,
callback_sfn: ThreadsafeFunction<String>,
) -> Result<(), String> {
) -> Result<()> {
let backend: &mut Box<dyn VersionBackend + Send> =
droplet_handler.create_backend_for_path(dir).ok_or(napi::Error::from_reason("Could not create backend for path."))?;
let backend: &'static mut Box<dyn VersionBackend + Send> =
unsafe { std::mem::transmute(backend) };
thread::spawn(move || {
let base_dir = Path::new(&dir);
let mut backend = create_backend_for_path(base_dir).unwrap();
let files = backend.list_files();
// Filepath to chunk data
@@ -54,7 +56,7 @@ pub fn generate_manifest(
let mut i: i32 = 0;
for version_file in files {
let raw_reader= backend.reader(&version_file).unwrap();
let raw_reader = backend.reader(&version_file).unwrap();
let mut reader = BufReader::with_capacity(CHUNK_SIZE, raw_reader);
let mut chunk_data = ChunkData {
@@ -106,4 +108,4 @@ pub fn generate_manifest(
});
Ok(())
}
}
+101 -73
View File
@@ -1,24 +1,34 @@
use core::arch;
#[cfg(unix)]
use std::os::unix::fs::PermissionsExt;
use std::{
fs::File,
io::{self, Read, Seek},
path::PathBuf,
pin::Pin,
rc::Rc,
fs::{self, metadata, File},
io::{self, Read, Sink},
path::{Path, PathBuf},
sync::Arc,
};
use flate2::read::DeflateDecoder;
use rawzip::{
FileReader, ReaderAt, ZipArchive, ZipArchiveEntryWayfinder, ZipEntry, RECOMMENDED_BUFFER_SIZE,
FileReader, ZipArchive, ZipArchiveEntryWayfinder, ZipEntry, ZipReader, RECOMMENDED_BUFFER_SIZE,
};
use crate::version::{
types::{MinimumFileObject, Skippable, VersionBackend, VersionFile},
utils::_list_files,
};
use crate::version::types::{MinimumFileObject, Skippable, VersionBackend, VersionFile};
pub fn _list_files(vec: &mut Vec<PathBuf>, path: &Path) {
if metadata(path).unwrap().is_dir() {
let paths = fs::read_dir(path).unwrap();
for path_result in paths {
let full_path = path_result.unwrap().path();
if metadata(&full_path).unwrap().is_dir() {
_list_files(vec, &full_path);
} else {
vec.push(full_path);
}
}
}
}
#[derive(Clone)]
pub struct PathVersionBackend {
pub base_dir: PathBuf,
}
@@ -30,40 +40,55 @@ impl VersionBackend for PathVersionBackend {
let mut results = Vec::new();
for pathbuf in vec.iter() {
let file = File::open(pathbuf.clone()).unwrap();
let relative = pathbuf.strip_prefix(self.base_dir.clone()).unwrap();
let metadata = file.try_clone().unwrap().metadata().unwrap();
let permission_object = metadata.permissions();
let permissions = {
let perm: u32;
#[cfg(target_family = "unix")]
{
perm = permission_object.mode();
}
#[cfg(not(target_family = "unix"))]
{
perm = 0
}
perm
};
results.push(VersionFile {
relative_filename: relative.to_string_lossy().to_string(),
permission: permissions,
size: metadata.len(),
});
results.push(
self
.peek_file(relative.to_str().unwrap().to_owned())
.unwrap(),
);
}
results
}
fn reader(&mut self, file: &VersionFile) -> Option<Box<(dyn MinimumFileObject + 'static)>> {
fn reader(&mut self, file: &VersionFile) -> Option<Box<dyn MinimumFileObject + 'static>> {
let file = File::open(self.base_dir.join(file.relative_filename.clone())).ok()?;
return Some(Box::new(file));
}
fn peek_file(&mut self, sub_path: String) -> Option<VersionFile> {
let pathbuf = self.base_dir.join(sub_path.clone());
if !pathbuf.exists() {
return None;
};
let file = File::open(pathbuf.clone()).unwrap();
let metadata = file.try_clone().unwrap().metadata().unwrap();
let permission_object = metadata.permissions();
let permissions = {
let perm: u32;
#[cfg(target_family = "unix")]
{
perm = permission_object.mode();
}
#[cfg(not(target_family = "unix"))]
{
perm = 0
}
perm
};
Some(VersionFile {
relative_filename: sub_path,
permission: permissions,
size: metadata.len(),
})
}
}
#[derive(Clone)]
pub struct ZipVersionBackend {
archive: Arc<ZipArchive<FileReader>>,
}
@@ -75,46 +100,51 @@ impl ZipVersionBackend {
}
}
pub fn new_entry(
pub fn new_entry<'archive>(
&self,
entry: ZipEntry<'_, FileReader>,
wayfinder: ZipArchiveEntryWayfinder,
) -> ZipFileWrapper {
let (offset, end_offset) = entry.compressed_data_range();
ZipFileWrapper {
archive: self.archive.clone(),
wayfinder,
offset,
end_offset,
}
entry: ZipEntry<'archive, FileReader>,
) -> ZipFileWrapper<'archive> {
let deflater = DeflateDecoder::new(entry.reader());
ZipFileWrapper { reader: deflater }
}
}
pub struct ZipFileWrapper {
pub archive: Arc<ZipArchive<FileReader>>,
wayfinder: ZipArchiveEntryWayfinder,
offset: u64,
end_offset: u64,
pub struct ZipFileWrapper<'archive> {
reader: DeflateDecoder<ZipReader<'archive, FileReader>>,
}
impl Read for ZipFileWrapper {
impl<'a> Read for ZipFileWrapper<'a> {
fn read(&mut self, buf: &mut [u8]) -> std::io::Result<usize> {
let read_size = buf.len().min((self.end_offset - self.offset) as usize);
let read = self
.archive
.get_ref()
.read_at(&mut buf[..read_size], self.offset)?;
self.offset += read as u64;
let read = self.reader.read(buf)?;
Ok(read)
}
}
impl Skippable for ZipFileWrapper {
impl<'a> Skippable for ZipFileWrapper<'a> {
fn skip(&mut self, amount: u64) {
self.offset += amount;
io::copy(&mut self.reader.by_ref().take(amount), &mut Sink::default()).unwrap();
}
}
impl MinimumFileObject for ZipFileWrapper {}
impl<'a> MinimumFileObject for ZipFileWrapper<'a> {}
impl ZipVersionBackend {
fn find_wayfinder(&mut self, filename: &str) -> Option<ZipArchiveEntryWayfinder> {
let read_buffer = &mut [0u8; RECOMMENDED_BUFFER_SIZE];
let mut entries = self.archive.entries(read_buffer);
let entry = loop {
if let Some(v) = entries.next_entry().unwrap() {
if v.file_path().try_normalize().unwrap().as_ref() == filename {
break Some(v);
}
} else {
break None;
}
}?;
let wayfinder = entry.wayfinder();
Some(wayfinder)
}
}
impl VersionBackend for ZipVersionBackend {
fn list_files(&mut self) -> Vec<VersionFile> {
let mut results = Vec::new();
@@ -133,24 +163,22 @@ impl VersionBackend for ZipVersionBackend {
results
}
fn reader(&mut self, file: &VersionFile) -> Option<Box<(dyn MinimumFileObject)>> {
let read_buffer = &mut [0u8; RECOMMENDED_BUFFER_SIZE];
let mut entries = self.archive.entries(read_buffer);
let entry = loop {
if let Some(v) = entries.next_entry().unwrap() {
if v.file_path().try_normalize().unwrap().as_ref() == &file.relative_filename {
break Some(v);
}
} else {
break None;
}
}?;
let wayfinder = entry.wayfinder();
fn reader(&mut self, file: &VersionFile) -> Option<Box<dyn MinimumFileObject + '_>> {
let wayfinder = self.find_wayfinder(&file.relative_filename)?;
let local_entry = self.archive.get_entry(wayfinder).unwrap();
let wrapper = self.new_entry(local_entry, wayfinder);
let wrapper = self.new_entry(local_entry);
Some(Box::new(wrapper))
}
fn peek_file(&mut self, sub_path: String) -> Option<VersionFile> {
let entry = self.find_wayfinder(&sub_path)?;
Some(VersionFile {
relative_filename: sub_path,
permission: 0,
size: entry.uncompressed_size_hint(),
})
}
}
+12 -8
View File
@@ -1,7 +1,9 @@
use std::{
fmt::Debug, io::{Read, Seek, SeekFrom}
fmt::Debug,
io::{Read, Seek, SeekFrom},
};
use dyn_clone::DynClone;
use tokio::io::{self, AsyncRead};
#[derive(Debug, Clone)]
@@ -27,12 +29,11 @@ pub trait MinimumFileObject: Read + Send + Skippable {}
impl<T: Read + Send + Seek> MinimumFileObject for T {}
// Intentionally not a generic, because of types in read_file
pub struct ReadToAsyncRead {
pub inner: Box<(dyn Read + Send)>,
pub backend: Box<(dyn VersionBackend + Send)>,
pub struct ReadToAsyncRead<'a> {
pub inner: Box<dyn Read + Send + 'a>,
}
impl AsyncRead for ReadToAsyncRead {
impl<'a> AsyncRead for ReadToAsyncRead<'a> {
fn poll_read(
mut self: std::pin::Pin<&mut Self>,
_cx: &mut std::task::Context<'_>,
@@ -40,13 +41,16 @@ impl AsyncRead for ReadToAsyncRead {
) -> std::task::Poll<io::Result<()>> {
let mut read_buf = [0u8; 8192];
let var_name = self.inner.read(&mut read_buf).unwrap();
let amount = var_name;
let amount = var_name.min(buf.remaining());
buf.put_slice(&read_buf[0..amount]);
std::task::Poll::Ready(Ok(()))
}
}
pub trait VersionBackend {
pub trait VersionBackend: DynClone {
fn list_files(&mut self) -> Vec<VersionFile>;
fn reader(&mut self, file: &VersionFile) -> Option<Box<(dyn MinimumFileObject)>>;
fn peek_file(&mut self, sub_path: String) -> Option<VersionFile>;
fn reader(&mut self, file: &VersionFile) -> Option<Box<dyn MinimumFileObject + '_>>;
}
dyn_clone::clone_trait_object!(VersionBackend);
+136 -100
View File
@@ -1,10 +1,6 @@
use std::{
fs::{self, metadata, File},
io::Read,
path::{Path, PathBuf},
};
use std::{collections::HashMap, fs::File, io::Read, path::Path};
use napi::{bindgen_prelude::*, tokio_stream::StreamExt};
use napi::{bindgen_prelude::*, sys::napi_value__, tokio_stream::StreamExt};
use tokio_util::codec::{BytesCodec, FramedRead};
use crate::version::{
@@ -12,120 +8,160 @@ use crate::version::{
types::{ReadToAsyncRead, VersionBackend, VersionFile},
};
pub fn _list_files(vec: &mut Vec<PathBuf>, path: &Path) {
if metadata(path).unwrap().is_dir() {
let paths = fs::read_dir(path).unwrap();
for path_result in paths {
let full_path = path_result.unwrap().path();
if metadata(&full_path).unwrap().is_dir() {
_list_files(vec, &full_path);
} else {
vec.push(full_path);
}
}
/**
* Append new backends here
*/
pub fn create_backend_constructor<'a>(
path: &Path,
) -> Option<Box<dyn FnOnce() -> Box<dyn VersionBackend + Send + 'a>>> {
if !path.exists() {
return None;
}
}
pub fn create_backend_for_path<'a>(path: &Path) -> Option<Box<(dyn VersionBackend + Send + 'a)>> {
let is_directory = path.is_dir();
if is_directory {
return Some(Box::new(PathVersionBackend {
base_dir: path.to_path_buf(),
}));
let base_dir = path.to_path_buf();
return Some(Box::new(move || Box::new(PathVersionBackend { base_dir })));
};
if path.to_string_lossy().ends_with(".zip") {
let f = File::open(path.to_path_buf()).unwrap();
return Some(Box::new(ZipVersionBackend::new(f)));
return Some(Box::new(|| Box::new(ZipVersionBackend::new(f))));
}
None
}
#[napi]
pub fn has_backend_for_path(path: String) -> bool {
let path = Path::new(&path);
let has_backend = create_backend_for_path(path).is_some();
has_backend
}
#[napi]
pub fn list_files(path: String) -> Result<Vec<String>> {
let path = Path::new(&path);
let mut backend =
create_backend_for_path(path).ok_or(napi::Error::from_reason("No backend for path"))?;
let files = backend.list_files();
Ok(files.into_iter().map(|e| e.relative_filename).collect())
}
/**
* This is inefficient, but is used in attempt to keep the interface simple
* Persistent object so we can cache things between commands
*/
#[napi]
pub fn peek_file(path: String, sub_path: String) -> Result<u64> {
let path = Path::new(&path);
let mut backend =
create_backend_for_path(path).ok_or(napi::Error::from_reason("No backend for path"))?;
let files = backend.list_files();
let file = files
.iter()
.find(|e| e.relative_filename == sub_path)
.ok_or(napi::Error::from_reason("Can't find file to peek"))?;
return Ok(file.size.try_into().unwrap());
#[napi(js_name = "DropletHandler")]
pub struct DropletHandler<'a> {
backend_cache: HashMap<String, Box<dyn VersionBackend + Send + 'a>>,
}
#[napi]
pub fn read_file(
path: String,
sub_path: String,
env: &Env,
start: Option<BigInt>,
end: Option<BigInt>,
) -> Option<ReadableStream<'_, BufferSlice<'_>>> {
let path = Path::new(&path);
let mut backend = create_backend_for_path(path).unwrap();
let version_file = VersionFile {
relative_filename: sub_path,
permission: 0, // Shouldn't matter
size: 0, // Shouldn't matter
};
// Use `?` operator for cleaner error propagation from `Option`
let mut reader = backend.reader(&version_file)?;
// Skip the 'start' amount of bytes without seek
if let Some(skip) = start.clone() {
reader.skip(skip.get_u64().1.into());
// io::copy(&mut reader.by_ref().take(skip.into()), &mut io::sink()).unwrap();
impl<'a> DropletHandler<'a> {
#[napi(constructor)]
pub fn new() -> Self {
DropletHandler {
backend_cache: HashMap::new(),
}
}
let async_reader = if let Some(limit) = end {
let amount = limit.get_u64().1 - start.map_or(Some(0), |v| Some(v.get_u64().1)).unwrap();
ReadToAsyncRead {
inner: Box::new(reader.take(amount.into())),
backend,
}
} else {
ReadToAsyncRead {
inner: reader,
backend,
}
};
pub fn create_backend_for_path(
&mut self,
path: String,
) -> Option<&mut Box<dyn VersionBackend + Send + 'a>> {
let fs_path = Path::new(&path);
let constructor = create_backend_constructor(fs_path)?;
// Create a FramedRead stream with BytesCodec for chunking
let stream = FramedRead::new(async_reader, BytesCodec::new())
// Use StreamExt::map to transform each Result item
.map(|result_item| {
result_item
// Apply Result::map to transform Ok(BytesMut) to Ok(Vec<u8>)
.map(|bytes| bytes.to_vec())
// Apply Result::map_err to transform Err(std::io::Error) to Err(napi::Error)
.map_err(|e| napi::Error::from(e)) // napi::Error implements From<tokio::io::Error>
let existing_backend = self.backend_cache.entry(path).or_insert_with(|| {
let backend = constructor();
backend
});
// Create the napi-rs ReadableStream from the tokio_stream::Stream
// The unwrap() here means if stream creation fails, it will panic.
// For a production system, consider returning Result<Option<...>> and handling this.
Some(ReadableStream::create_with_stream_bytes(env, stream).unwrap())
Some(existing_backend)
}
#[napi]
pub fn has_backend_for_path(&self, path: String) -> bool {
let path = Path::new(&path);
let has_backend = create_backend_constructor(path).is_some();
has_backend
}
#[napi]
pub fn list_files(&mut self, path: String) -> Result<Vec<String>> {
let backend = self
.create_backend_for_path(path)
.ok_or(napi::Error::from_reason("No backend for path"))?;
let files = backend.list_files();
Ok(files.into_iter().map(|e| e.relative_filename).collect())
}
#[napi]
pub fn peek_file(&mut self, path: String, sub_path: String) -> Result<u64> {
let backend = self
.create_backend_for_path(path)
.ok_or(napi::Error::from_reason("No backend for path"))?;
let file = backend
.peek_file(sub_path)
.ok_or(napi::Error::from_reason("Can't find file to peek"))?;
return Ok(file.size.try_into().unwrap());
}
#[napi]
pub fn read_file(
&mut self,
reference: Reference<DropletHandler<'static>>,
path: String,
sub_path: String,
env: Env,
start: Option<BigInt>,
end: Option<BigInt>,
) -> Result<JsDropStreamable> {
let stream = reference.share_with(env, |handler| {
let backend = handler
.create_backend_for_path(path)
.ok_or(napi::Error::from_reason("Failed to create backend."))?;
let version_file = VersionFile {
relative_filename: sub_path,
permission: 0, // Shouldn't matter
size: 0, // Shouldn't matter
};
// Use `?` operator for cleaner error propagation from `Option`
let mut reader = backend.reader(&version_file).ok_or(napi::Error::from_reason("Failed to create reader."))?;
if let Some(skip) = start.clone() {
reader.skip(skip.get_u64().1.into());
// io::copy(&mut reader.by_ref().take(skip.into()), &mut io::sink()).unwrap();
}
let async_reader = if let Some(limit) = end {
let amount = limit.get_u64().1 - start.map_or(Some(0), |v| Some(v.get_u64().1)).unwrap();
ReadToAsyncRead {
inner: Box::new(reader.take(amount.into())),
}
} else {
ReadToAsyncRead { inner: reader }
};
// Create a FramedRead stream with BytesCodec for chunking
let stream = FramedRead::new(async_reader, BytesCodec::new())
// Use StreamExt::map to transform each Result item
.map(|result_item| {
result_item
// Apply Result::map to transform Ok(BytesMut) to Ok(Vec<u8>)
.map(|bytes| bytes.to_vec())
// Apply Result::map_err to transform Err(std::io::Error) to Err(napi::Error)
.map_err(|e| napi::Error::from(e)) // napi::Error implements From<tokio::io::Error>
});
// Create the napi-rs ReadableStream from the tokio_stream::Stream
// The unwrap() here means if stream creation fails, it will panic.
// For a production system, consider returning Result<Option<...>> and handling this.
Ok(ReadableStream::create_with_stream_bytes(&env, stream).unwrap())
})?;
Ok(JsDropStreamable {
inner: stream,
})
}
}
#[napi]
pub struct JsDropStreamable {
inner: SharedReference<DropletHandler<'static>, ReadableStream<'static, BufferSlice<'static>>>,
}
#[napi]
impl JsDropStreamable {
#[napi]
pub fn get_stream(&self) -> *mut napi_value__ {
self.inner.raw()
}
}