chgk_ledb/src/main.rs

extern crate serde;
extern crate serde_derive;
#[macro_use]
extern crate serde_json;
extern crate ledb;
extern crate ledb_types;
extern crate zip;

use clap::{Parser, Subcommand};
use rand::seq::IteratorRandom;

use std::io::{Read, Seek, Write};
use std::path::PathBuf;
use std::time::Instant;
use std::{fs, io, sync::mpsc, thread};

use ledb::{Options, Storage};

mod questions;
mod source;

use crate::questions::{Question, QuestionsConverter};
use crate::source::ReadSourceQuestionsBatches;

const ZIP_FILENAME: &str = "json.zip";
const DB_DIR: &str = "db";

#[derive(Subcommand, Debug)]
enum Command {
    Write,
    Compact,
    Print {
        #[clap(value_parser, default_value = "0")]
        id: u32,
    },
    ZipPrint {
        #[clap(value_parser, default_value = "0")]
        file_num: usize,
        #[clap(value_parser, default_value = "0")]
        num: usize,
    },
    Write2,
    Print2 {
        #[clap(value_parser, default_value = "0")]
        id: u32,
    },
}

#[derive(Parser, Debug)]
#[clap(author, version, about, long_about = None)]
#[clap(propagate_version = true)]
struct Cli {
    #[clap(subcommand)]
    command: Command,
    #[clap(short, long, action)]
    measure: bool,
}

fn reader_task(tx: mpsc::Sender<Question>) {
    let zip_file = fs::File::open(ZIP_FILENAME).unwrap();
    let zip_reader = io::BufReader::new(zip_file);
    let archive = zip::ZipArchive::new(zip_reader).unwrap();
    let mut source_questions = archive.source_questions();
    for question in source_questions.convert() {
        let res = tx.send(question);
        if res.is_err() {
            break;
        }
    }
    println!("read done");
}
fn db_writer_task(rx: mpsc::Receiver<Question>) {
    let out_file: PathBuf = [DB_DIR, "data.mdb"].into_iter().collect();
    match fs::metadata(&out_file) {
        Ok(x) if x.is_file() => {
            fs::remove_file(&out_file).unwrap();
            println!(r#""{}" removed"#, out_file.to_str().unwrap());
        }
        _ => {}
    };

    let options: Options = serde_json::from_value(json!({
        "map_size": 900 * 1024 * 1024, // 900mb
        "write_map": true,
        "map_async": true,
        "no_lock": true,
        "no_meta_sync": true,
        "no_sync": true,
    }))
    .unwrap();

    let storage = Storage::new(DB_DIR, options).unwrap();
    let collection = storage.collection("questions").unwrap();

    let mut count: usize = 0;
    let count = &mut count;
    rx.into_iter().for_each(|question| {
        let result = collection.insert(&question);
        if result.is_err() {
            println!("-- {:#?}", question);
            panic!("{:#?}", result);
        } else {
            *count += 1;
        }
    });

    println!("inserted {}\nwriting...", count);
    storage.sync(true).unwrap();
    print!("stats: ");
    let stats = storage.stat().unwrap();
    println!("{:?}", stats);
    drop(storage);
    println!("write done");
}

fn write_db() {
    let (tx, rx) = mpsc::channel::<Question>();
    [
        thread::spawn(move || reader_task(tx)),
        thread::spawn(move || db_writer_task(rx)),
    ]
    .into_iter()
    .for_each(|handle| handle.join().expect("thread panic"));
    println!("all done");
}

fn print_question_from<F>(get_q: F)
where
    F: FnOnce() -> Option<Question>,
{
    let q = get_q().unwrap();
    println!("{:#?}", q)
}

fn read_from_zip(file_num: usize, mut num: usize) -> Option<Question> {
    let mut rng = rand::thread_rng();
    let zip_file = fs::File::open(ZIP_FILENAME).unwrap();
    let zip_reader = io::BufReader::new(zip_file);
    let archive = zip::ZipArchive::new(zip_reader).unwrap();

    let mut source_questions = archive.source_questions();
    let (filename, batch) = if file_num == 0 {
        source_questions.choose(&mut rng).unwrap()
    } else {
        source_questions.nth(file_num - 1).unwrap()
    };
    let mut batch = batch.unwrap();
    batch.filename = filename;
    let questions: Vec<Question> = batch.into();
    if num == 0 {
        num = (1..=questions.len()).choose(&mut rng).unwrap();
    }
    Some(questions[num - 1].clone())
}

fn compact_db() {
    let options: Options = serde_json::from_value(json!({
        "write_map": true,
        "map_async": true,
        "no_lock": true,
        "no_meta_sync": true,
        "no_sync": true,
        "compact": true,
    }))
    .unwrap();

    let storage = Storage::new(DB_DIR, options).unwrap();

    storage.sync(true).unwrap();
    let stats = storage.stat().unwrap();
    println!("{:?}", stats);
    drop(storage);
}

fn read_from_db(mut id: u32) -> Option<Question> {
    let options: Options = serde_json::from_value(json!({
        "read_only": true,
        "map_async": true,
        "no_lock": true,
    }))
    .unwrap();

    let storage = Storage::new(DB_DIR, options).unwrap();
    let collection = storage.collection("questions").unwrap();
    let mut rng = rand::thread_rng();

    if id == 0 {
        let last_id = collection.last_id().unwrap();
        id = (1..=last_id).choose(&mut rng).unwrap();
    }

    collection.get::<Question>(id).unwrap()
}

// measure and return time elapsed in `func` in seconds
pub fn measure<F: FnOnce()>(func: F) -> f64 {
    let start = Instant::now();
    func();
    let elapsed = start.elapsed();
    (elapsed.as_secs() as f64) + (elapsed.subsec_nanos() as f64 / 1_000_000_000.0)
}

pub fn measure_and_print<F: FnOnce()>(func: F) {
    let m = measure(func);
    eprintln!("{}", m);
}

fn main() {
    let args = Cli::parse();

    let mut action: Box<dyn FnOnce()> = match &args.command {
        Command::Write => Box::new(write_db),
        Command::Compact => Box::new(compact_db),
        Command::Print { id } => {
            let get_question = Box::new(|| read_from_db(*id));
            Box::new(|| print_question_from(get_question))
        }
        Command::ZipPrint { file_num, num } => {
            let get_question = Box::new(|| read_from_zip(*file_num, *num));
            Box::new(|| print_question_from(get_question))
        }
        Command::Write2 => Box::new(write_db2),
        Command::Print2 { id } => {
            let get_question = Box::new(|| read_from_db2(*id));
            Box::new(|| print_question_from(get_question))
        }
    };

    if args.measure {
        action = Box::new(|| measure_and_print(action));
    }

    action();
}

fn read_from_db2(id: u32) -> Option<Question> {
    const LEN_SIZE: usize = std::mem::size_of::<u32>();
    let cfg = bincode::config::standard();

    let input = fs::File::open("test.bin").expect("open input");
    let mut input = std::io::BufReader::with_capacity(100 * 1024 * 1024, input);

    let mut len_data: [u8; LEN_SIZE] = [0; LEN_SIZE];
    input.read_exact(&mut len_data).expect("read len");
    let tab_len = u32::from_le_bytes(len_data) as usize;
    let records_count = tab_len - 1;

    // println!(
    //     "read tab_len done: {}, pos {}",
    //     tab_len,
    //     input.stream_position().unwrap()
    // );

    let index = match id {
        0 => {
            let mut rng = rand::thread_rng();
            (1..records_count).into_iter().choose(&mut rng).unwrap()
        }
        _ => (id - 1) as usize,
    };

    // println!("index {}", index);
    assert!(index < records_count);

    let tab_tail_len = (tab_len - (index + 2)) * LEN_SIZE;

    let mut pos_curr_data: [u8; LEN_SIZE] = [0; LEN_SIZE];
    let mut pos_next_data: [u8; LEN_SIZE] = [0; LEN_SIZE];
    input
        .seek_relative((index * LEN_SIZE).try_into().expect("index to i64"))
        .expect("seek to tab pos");
    input
        .read_exact(&mut pos_curr_data)
        .expect("read current pos");
    input.read_exact(&mut pos_next_data).expect("read next pos");
    
    let data_pos = u32::from_le_bytes(pos_curr_data);
    let data_pos_next = u32::from_le_bytes(pos_next_data);
    let data_len = data_pos_next - data_pos;

    // println!(
    //     "pos {} | next {} | len {} | tab_tail_len {}",
    //     data_pos, data_pos_next, data_len, tab_tail_len
    // );
    let data_pos = data_pos + tab_tail_len as u32;

    input.seek_relative(data_pos as i64).expect("q seek");
    let reader = input.take(data_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;

    // println!("read done");

    Some(question)
}
fn write_db2() {
    let (tx, rx) = mpsc::channel::<Question>();
    [
        thread::spawn(move || reader_task(tx)),
        thread::spawn(move || db_writer2_task(rx)),
    ]
    .into_iter()
    .for_each(|handle| handle.join().expect("thread panic"));
    println!("all done");
}
fn db_writer2_task(rx: mpsc::Receiver<Question>) {
    const LEN_SIZE: usize = std::mem::size_of::<u32>();
    const COMP_DATA_LEVEL: i32 = 2;

    let cfg = bincode::config::standard();
    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;
    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), COMP_DATA_LEVEL).expect("zstd q encode");
        let len = buf.write(&data).expect("write question");
        table.push(pos);

        //println!("write [{}]: {}", num, pos);

        pos += len as u32;
        num += 1;
    });
    table.push(pos);
    println!("write [{}]: {}", num, pos);

    println!(
        "zbuf done, tab len {}, buf size {}",
        table.len(),
        buf.position()
    );

    buf.set_position(0);

    let tab_data = vec![0u8; (table.len() + 1) * LEN_SIZE];
    let mut tab_cursor = Cursor::new(tab_data);
    let len_data = (table.len() as u32).to_le_bytes();
    tab_cursor.write_all(&len_data).expect("write len");
    for pos in table {
        let pos_data = pos.to_le_bytes();
        tab_cursor.write_all(&pos_data).expect("write pos");
    }

    println!("tab buf done, len = {}", tab_cursor.position());
    tab_cursor.set_position(0);

    let out = fs::File::create("test.bin").expect("out create");
    let mut out = std::io::BufWriter::with_capacity(500 * 1024 * 1024, out);
    std::io::copy(&mut tab_cursor, &mut out).expect("write tab");

    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");

    println!("write done");
}