min tab 1

This commit is contained in:
Dmitry Belyaev 2022-10-07 11:53:17 +03:00
parent 5a4086cfdf
commit 42fc486872
Signed by: b4tman
GPG Key ID: 41A00BF15EA7E5F3
1 changed files with 44 additions and 24 deletions

View File

@ -9,7 +9,7 @@ extern crate zip;
use clap::{Parser, Subcommand};
use rand::seq::IteratorRandom;
use std::io::{Read, Write};
use std::io::{Read, Seek, Write};
use std::path::PathBuf;
use std::time::Instant;
use std::{fs, io, sync::mpsc, thread};
@ -230,12 +230,6 @@ fn main() {
action();
}
#[derive(bincode::Decode, bincode::Encode)]
struct VEntry {
pos: u32,
len: u16,
}
fn read_from_db2(id: u32) -> Option<Question> {
use std::io::Cursor;
const LEN_SIZE: usize = std::mem::size_of::<u64>();
@ -253,17 +247,33 @@ fn read_from_db2(id: u32) -> Option<Question> {
let mut zdata = vec![0u8; len];
input.read_exact(&mut zdata[..len]).expect("read ztab");
let tab_data = zstd::decode_all(Cursor::new(zdata)).expect("zstd decode table");
let tab: (Vec<VEntry>, usize) =
let tab: (Vec<u32>, usize) =
bincode::decode_from_slice(&tab_data, cfg).expect("bincode decode tab");
let tab = tab.0;
//println!("read tab done");
//println!("read tab done, len {}", tab.len());
let index = match id {
0 => {
let mut rng = rand::thread_rng();
(1..tab.len()).into_iter().choose(&mut rng).unwrap()
}
_ => (id - 1) as usize,
};
let choosen = tab.get(id as usize).expect("get by id");
input.seek_relative(choosen.pos as i64).expect("q seek");
let reader = input.take(choosen.len as u64);
//println!("index {}", index);
let pos = *tab.get(index).expect("get pos");
let pos_next = *tab.get((index + 1) as usize).expect("get pos next");
let len = pos_next - pos;
//println!("pos {} | next {} | len {}", pos, pos_next, len);
input.seek_relative(pos as i64).expect("q seek");
let reader = input.take(len as u64);
let data = zstd::decode_all(reader).expect("zstd decode data");
//println!("zstd decoded len {}", data.len());
let question: (Question, usize) =
bincode::decode_from_slice(&data, cfg).expect("bincode decode q");
let question = question.0;
@ -284,35 +294,42 @@ fn write_db2() {
}
fn db_writer2_task(rx: mpsc::Receiver<Question>) {
const LEN_SIZE: usize = std::mem::size_of::<u64>();
const COMP_DATA_LEVEL: i32 = 2;
const COMP_HDR_LEVEL: i32 = 2;
let cfg = bincode::config::standard().with_fixed_int_encoding();
let mut table: Vec<VEntry> = vec![];
let mut table: Vec<u32> = vec![];
let buf_data: Vec<u8> = Vec::with_capacity(500 * 1024 * 1024);
use std::io::Cursor;
let mut buf = Cursor::new(buf_data);
let mut pos: u32 = 0;
rx.into_iter().for_each(|q| {
let mut num = 1;
rx.into_iter().for_each(|mut q| {
q.num = num;
let data = bincode::encode_to_vec(q, cfg).expect("bincode q encode");
let data = zstd::encode_all(Cursor::new(data), 3).expect("zstd q encode");
let data = zstd::encode_all(Cursor::new(data), COMP_DATA_LEVEL).expect("zstd q encode");
let len = buf.write(&data).expect("write question");
table.push(VEntry {
pos,
len: len.try_into().expect("len try_into len"),
});
table.push(pos);
pos += len as u32;
num += 1;
});
table.push(pos);
println!(
"zbuf done, tab len {}, buf size {}",
table.len(),
buf.position()
);
buf.set_position(0);
println!("zbuf done");
let tab_data = bincode::encode_to_vec(&table, cfg).expect("encode table");
let zdata = zstd::encode_all(Cursor::new(tab_data), 3).expect("zstd enc table");
let zdata = zstd::encode_all(Cursor::new(tab_data), COMP_HDR_LEVEL).expect("zstd enc table");
let zlen = zdata.len() as u64;
println!("z tab done");
println!("z tab done, tab data_len = {}", zlen);
let out = fs::File::create("test.bin").expect("out create");
let mut out = std::io::BufWriter::with_capacity(500 * 1024 * 1024, out);
@ -321,6 +338,9 @@ fn db_writer2_task(rx: mpsc::Receiver<Question>) {
let ztab_writed = out.write(&zdata).expect("write tab zdata");
assert_eq!(ztab_writed, zdata.len());
drop(zdata);
println!("header write done, pos: {}", out.stream_position().unwrap());
std::io::copy(&mut buf, &mut out).expect("copy z buf");
drop(buf);
out.flush().expect("out flush");