Fix delete bug

This commit is contained in:
Yuriy Dupyn 2024-02-03 21:34:50 +01:00
parent 3bf04ae2d6
commit a345bf99c6
4 changed files with 107 additions and 62 deletions

View file

@ -10,12 +10,13 @@ use bincode::{Decode, Encode};
use crate::binary_coding::encode; use crate::binary_coding::encode;
use crate::entry::{Entry, EntryDetailed}; use crate::entry::{Entry, EntryDetailed};
use crate::entry_header::EntryHeaderWithDataSize; use crate::entry_header::{EntryHeaderWithDataSize, EntryHeader};
use crate::store_header::StoreHeader; use crate::store_header::StoreHeader;
use crate::storage_engine::{Store, FilePosition, Result, ROWS_FILE_NAME}; use crate::storage_engine::{Store, FilePosition, Result, ROWS_FILE_NAME};
#[async_trait] #[async_trait]
trait PrimitiveCursor<T> { // TODO: Make this private
pub trait PrimitiveCursor<T> {
fn file(&mut self) -> &mut File; fn file(&mut self) -> &mut File;
fn eof_file_position(&self) -> FilePosition; fn eof_file_position(&self) -> FilePosition;
@ -30,22 +31,23 @@ trait PrimitiveCursor<T> {
Ok(result) Ok(result)
} }
async fn seek_to(&mut self, file_position: FilePosition) -> Result<()> { async fn seek_to(&mut self, file_position: FilePosition) -> Result<FilePosition> {
self.file().seek(SeekFrom::Start(file_position)).await?; let file_position = self.file().seek(SeekFrom::Start(file_position)).await?;
Ok(()) Ok(file_position)
} }
async fn seek_to_start(&mut self) -> Result<()> { // Start of the file i.e. the Header, not the entries.
self.file().seek(SeekFrom::Start(0)).await?; async fn seek_to_start(&mut self) -> Result<FilePosition> {
Ok(()) let file_position = self.file().seek(SeekFrom::Start(0)).await?;
Ok(file_position)
} }
async fn seek_to_end(&mut self) -> Result<()> { async fn seek_to_end(&mut self) -> Result<FilePosition> {
self.file().seek(SeekFrom::End(0)).await?; let file_position = self.file().seek(SeekFrom::End(0)).await?;
Ok(()) Ok(file_position)
} }
async fn seek_to_start_of_data(&mut self) -> Result<()> { async fn seek_to_start_of_data(&mut self) -> Result<FilePosition> {
self.seek_to(StoreHeader::SIZE as u64).await self.seek_to(StoreHeader::SIZE as u64).await
} }
@ -55,7 +57,10 @@ trait PrimitiveCursor<T> {
} }
async fn is_at_eof(&mut self) -> Result<bool> { async fn is_at_eof(&mut self) -> Result<bool> {
Ok(self.current_file_position().await? == self.eof_file_position()) let current_file_position = self.current_file_position().await?;
let eof_file_position = self.eof_file_position();
println!("IN EOF: current={}, eof_file_position={}", current_file_position, eof_file_position);
Ok(current_file_position == eof_file_position)
} }
} }
@ -90,9 +95,12 @@ pub trait CursorWithStoreHeader<T>: PrimitiveCursor<T> {
async fn next(&mut self) -> Result<Option<EntryDetailed<T>>> async fn next(&mut self) -> Result<Option<EntryDetailed<T>>>
where T: Decode where T: Decode
{ {
println!("are we at eof?");
if self.is_at_eof().await? { if self.is_at_eof().await? {
println!("YES");
return Ok(None) return Ok(None)
} }
println!("NO");
let entry_header = self.read_entry_header().await?; let entry_header = self.read_entry_header().await?;
@ -118,6 +126,7 @@ pub trait CursorWithStoreHeader<T>: PrimitiveCursor<T> {
async fn read_all_bytes(&mut self) -> std::result::Result<Vec<u8>, std::io::Error> { async fn read_all_bytes(&mut self) -> std::result::Result<Vec<u8>, std::io::Error> {
let mut bytes: Vec<u8> = vec![]; let mut bytes: Vec<u8> = vec![];
self.seek_to_start().await.map_err(|e| e.to_io_or_panic())?;
self.file().read_to_end(&mut bytes).await?; self.file().read_to_end(&mut bytes).await?;
Ok(bytes) Ok(bytes)
} }
@ -191,8 +200,11 @@ impl <T> ReadCursor<T> {
file, file,
data_type: store.data_type, data_type: store.data_type,
eof_file_position: 0, // This will be overwriten by the seek_to_start_of_data eof_file_position: 0,
}; };
let eof_file_position: FilePosition = cursor.seek_to_end().await?;
cursor.eof_file_position = eof_file_position;
cursor.seek_to_start_of_data().await?; cursor.seek_to_start_of_data().await?;
Ok(cursor) Ok(cursor)
@ -221,8 +233,11 @@ impl <'cursor, T> WriteCursor<'cursor, T> {
file, file,
data_type: store.data_type, data_type: store.data_type,
eof_file_position: 0, // This will be overwriten by the seek_to_start_of_data eof_file_position: 0,
}; };
let eof_file_position: FilePosition = cursor.seek_to_end().await?;
cursor.eof_file_position = eof_file_position;
cursor.seek_to_start_of_data().await?; cursor.seek_to_start_of_data().await?;
Ok(cursor) Ok(cursor)
@ -255,12 +270,10 @@ impl <'cursor, T> WriteCursor<'cursor, T> {
} }
// ===Entry Header Manipulation=== // ===Entry Header Manipulation===
// assumes we are at the start of the valid entry. // assumes we are at the start of valid entry.
async fn set_entry_is_deleted_to(&mut self, is_deleted: bool) -> Result<()> async fn set_new_entry_header(&mut self, entry_header: EntryHeader) -> Result<()> {
where T: Send let bytes: Vec<u8> = entry_header.encode()?;
{ self.write_bytes(&bytes).await?;
self.seek_to(EntryHeaderWithDataSize::IS_DELETED_OFFSET as u64).await?;
self.write_bytes(&encode::<bool>(&is_deleted)?).await?;
Ok(()) Ok(())
} }
@ -274,8 +287,7 @@ impl <'cursor, T> WriteCursor<'cursor, T> {
self.increment_total_count().await?; self.increment_total_count().await?;
let encoded_entry: Vec<u8> = entry.encode()?; let encoded_entry: Vec<u8> = entry.encode()?;
self.seek_to_end().await?; let file_position = self.seek_to_end().await?;
let file_position: FilePosition = self.current_file_position().await?;
self.write_bytes(&encoded_entry).await?; self.write_bytes(&encoded_entry).await?;
let eof_file_position: FilePosition = self.current_file_position().await?; let eof_file_position: FilePosition = self.current_file_position().await?;
@ -289,13 +301,19 @@ impl <'cursor, T> WriteCursor<'cursor, T> {
where T: Send where T: Send
{ {
self.seek_to(file_position).await?; self.seek_to(file_position).await?;
let entry_header = self.read_entry_header().await?; println!("Attempting to read the entry");
let mut entry_header = self.read_entry_header().await?;
println!("Entry Header == {:?}", entry_header);
if entry_header.is_deleted { if entry_header.is_deleted {
println!("Already deleted");
Ok(()) Ok(())
} else { } else {
println!("Marking as deleted");
self.increment_deleted_count().await?; self.increment_deleted_count().await?;
self.seek_to(file_position).await?; self.seek_to(file_position).await?;
self.set_entry_is_deleted_to(true).await?;
entry_header.is_deleted = true;
self.set_new_entry_header(entry_header.into()).await?;
self.attempt_garbage_collection_if_necessary().await?; self.attempt_garbage_collection_if_necessary().await?;
Ok(()) Ok(())

View file

@ -1,6 +1,6 @@
use bincode::{Decode, Encode}; use bincode::{Decode, Encode};
use crate::binary_coding::{encode, encode_sequence, encode_sequence_with_sizes, decode_sequence}; use crate::binary_coding::{encode_sequence, encode_sequence_with_sizes, decode_sequence};
use crate::storage_engine::Result; use crate::storage_engine::Result;
use crate::error::{Error, DecodeErrorKind}; use crate::error::{Error, DecodeErrorKind};
use crate::entry_header::{EntryHeader, EntryHeaderWithDataSize}; use crate::entry_header::{EntryHeader, EntryHeaderWithDataSize};
@ -17,13 +17,6 @@ pub struct EntryDetailed<T> {
data: Vec<T>, data: Vec<T>,
} }
impl EntryHeader {
fn encode(self: &EntryHeader) -> Result<Vec<u8>> {
let result: Vec<u8> = encode(&self.is_deleted)?;
Ok(result)
}
}
impl <T>Entry<T> { impl <T>Entry<T> {
pub fn new(data: Vec<T>) -> Self { pub fn new(data: Vec<T>) -> Self {
Self { header: EntryHeader { is_deleted: false }, data } Self { header: EntryHeader { is_deleted: false }, data }

View file

@ -1,4 +1,4 @@
use crate::binary_coding::{decode, decode_sequence}; use crate::binary_coding::{decode, encode, decode_sequence};
use crate::storage_engine::Result; use crate::storage_engine::Result;
use crate::error::{Error, DecodeErrorKind}; use crate::error::{Error, DecodeErrorKind};
use std::mem::size_of; use std::mem::size_of;
@ -14,6 +14,20 @@ pub struct EntryHeaderWithDataSize {
pub data_sizes: Vec<usize>, // vec![5, 6, 20] means that column 0 stores 5 bytes, column 1 stores 6 pub data_sizes: Vec<usize>, // vec![5, 6, 20] means that column 0 stores 5 bytes, column 1 stores 6
// bytes etc // bytes etc
} }
impl EntryHeader {
pub fn encode(self: &EntryHeader) -> Result<Vec<u8>> {
let result: Vec<u8> = encode(&self.is_deleted)?;
Ok(result)
}
}
impl From<EntryHeaderWithDataSize> for EntryHeader {
fn from(entry: EntryHeaderWithDataSize) -> Self {
Self { is_deleted: entry.is_deleted, }
}
}
impl EntryHeaderWithDataSize { impl EntryHeaderWithDataSize {
pub const IS_DELETED_OFFSET: usize = 0; pub const IS_DELETED_OFFSET: usize = 0;
pub const IS_DELETED_SIZE: usize = size_of::<bool>(); pub const IS_DELETED_SIZE: usize = size_of::<bool>();

View file

@ -9,7 +9,7 @@ mod store_header;
use crate::entry::{Entry, EntryDetailed}; use crate::entry::{Entry, EntryDetailed};
use crate::storage_engine::{Store, FilePosition}; use crate::storage_engine::{Store, FilePosition};
use crate::cursor::{ReadCursor, WriteCursor}; use crate::cursor::{ReadCursor, WriteCursor, CursorWithStoreHeader, PrimitiveCursor};
type Data = u32; type Data = u32;
@ -20,8 +20,8 @@ type Result<T> = std::result::Result<T, std::io::Error>;
async fn create_store() -> Result<Store<Data>> { async fn create_store() -> Result<Store<Data>> {
let mut store: Store<Data> = Store::new(TABLE_PATH, 5, 0).await.map_err(|e| e.to_io_or_panic())?; let mut store: Store<Data> = Store::new(TABLE_PATH, 5, 0).await.map_err(|e| e.to_io_or_panic())?;
println!("CREATED"); println!("CREATED");
println!("THE STORE: {:?}", store); // println!("THE STORE: {:?}", store);
println!("THE BYTES: {:?}", store.read_all_bytes().await?); // println!("THE BYTES: {:?}", store.read_all_bytes().await?);
Ok(store) Ok(store)
} }
@ -29,8 +29,8 @@ async fn create_store() -> Result<Store<Data>> {
async fn connect_store() -> Result<Store<Data>> { async fn connect_store() -> Result<Store<Data>> {
let mut store: Store<Data> = Store::connect(TABLE_PATH).await.map_err(|e| e.to_io_or_panic())?; let mut store: Store<Data> = Store::connect(TABLE_PATH).await.map_err(|e| e.to_io_or_panic())?;
println!("CONNECTED"); println!("CONNECTED");
println!("THE STORE: {:?}", store); // println!("THE STORE: {:?}", store);
println!("THE BYTES: {:?}", store.read_all_bytes().await?); // println!("THE BYTES: {:?}", store.read_all_bytes().await?);
Ok(store) Ok(store)
} }
@ -60,14 +60,7 @@ async fn read_entry(cursor: &mut ReadCursor<Data>, file_position: FilePosition)
todo!() todo!()
} }
async fn append_bunch_of_entries(store: &mut Store<Data>) -> Result<()> {
#[tokio::main]
async fn main() -> Result<()> {
println!("STOOOOOOOOOOOORAAAAAAAAAAAGE");
let mut store: Store<Data> = create_or_connect().await?;
{
let mut cursor = store.write_cursor().await.map_err(|e| e.to_io_or_panic())?; let mut cursor = store.write_cursor().await.map_err(|e| e.to_io_or_panic())?;
let entry0: Entry<u32> = Entry::new(vec![1, 2, 3, 4, 5]); let entry0: Entry<u32> = Entry::new(vec![1, 2, 3, 4, 5]);
append_entry(&mut cursor, &entry0).await?; append_entry(&mut cursor, &entry0).await?;
@ -81,28 +74,55 @@ async fn main() -> Result<()> {
let entry3: Entry<u32> = Entry::new(vec![50,50,50,50,50]); let entry3: Entry<u32> = Entry::new(vec![50,50,50,50,50]);
append_entry(&mut cursor, &entry3).await?; append_entry(&mut cursor, &entry3).await?;
Ok(())
}
#[tokio::main]
async fn main() -> Result<()> {
println!("STOOOOOOOOOOOORAAAAAAAAAAAGE");
let mut store: Store<Data> = create_or_connect().await?;
if store.header.total_count == 0 {
println!("INSERTING!");
append_bunch_of_entries(&mut store).await?;
}
{
let mut cursor = store.write_cursor().await.map_err(|e| e.to_io_or_panic())?;
let entry: Entry<u32> = Entry::new(vec![60, 50, 40, 30, 20]);
// let file_position = append_entry(&mut cursor, &entry).await?;
// let file_position = 215;
// cursor.seek_to(file_position).await.map_err(|e| e.to_io_or_panic())?;
// let entry_header = cursor.read_entry_header().await.map_err(|e| e.to_io_or_panic())?;
// println!("entry header = {:?}", entry_header);
// println!("FILE POSITION == {}", file_position);
// cursor.mark_deleted_at(file_position).await.map_err(|e| e.to_io_or_panic())?;
// let entry_header = cursor.read_entry_header().await.map_err(|e| e.to_io_or_panic())?;
// println!("entry header after delete = {:?}", entry_header);
} }
// println!("{:?}", store); // println!("{:?}", store);
// println!("{:?}", store.read_all_bytes().await?); // println!("{:?}", store.read_all_bytes().await?);
// let entry0: Entry<u32> = Entry::new(vec![99, 98, 97, 96, 95]); let mut cursor = store.read_cursor().await.map_err(|e| e.to_io_or_panic())?;
// append_entry(&mut cursor, &entry0).await?;
// let entry1: Entry<u32> = Entry::new(vec![50,50,50,50,50]); // let x = cursor.next().await.map_err(|e| e.to_io_or_panic())?;
// let file_position = append_entry(&mut cursor, &entry1).await?; // println!("{:?}", x);
// println!("CURRENT FILE_POSITION = {}", file_position); // let x = cursor.next().await.map_err(|e| e.to_io_or_panic())?;
// println!("{:?}", x);
// let x = cursor.next().await.map_err(|e| e.to_io_or_panic())?;
// println!("{:?}", x);
// let x = cursor.next().await.map_err(|e| e.to_io_or_panic())?;
// println!("{:?}", x);
// let x = cursor.next().await.map_err(|e| e.to_io_or_panic())?;
// println!("{:?}", x);
// Now file_position point to entry1. cursor.read_entries().await.map_err(|e| e.to_io_or_panic())?;
// cursor.mark_deleted_at(file_position).await.map_err(|e| e.to_io_or_panic())?;
// cursor.seek_to(file_position).await.map_err(|e| e.to_io_or_panic())?;
// let entry2: StoreEntry<u32> = StoreEntry::new_deleted(vec![3, 2, 1]);
// let cursor2 = store.append_entry(&entry2).await.map_err(|e| e.to_io_or_panic())?;
// println!("cursor2 = {}", cursor2);
println!("{:?}", store);
println!("{:?}", store.read_all_bytes().await?);
println!("DONE"); println!("DONE");