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;
use std::path::PathBuf;
use std::time::Instant;
use std::{fs, sync::mpsc, thread};

use ledb::{Options, Storage};

mod db;
mod questions;
mod source;

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

const ZIP_FILENAME: &str = "json.zip";
const NEW_DB_FILENAME: &str = "test.bin";
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 zip_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();

    let questions = source_questions
        .convert()
        .enumerate()
        .map(|(num, mut question)| {
            question.num = 1 + num as u32;
            question
        });
    for question in questions {
        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 count = collection.load(rx).expect("load");

    println!("loaded {count}");

    println!("syncing to disk...");
    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 || zip_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> {
    let mut reader: db::Reader<Question> =
        db::Reader::new(NEW_DB_FILENAME, 2048).expect("new db reader");

    let mut questions = reader.iter();

    match id {
        0 => {
            let mut rng = rand::thread_rng();
            questions.choose(&mut rng)
        }
        _ => questions.nth((id - 1) as usize),
    }
}
fn write_db2() {
    let (tx, rx) = mpsc::channel::<Question>();
    [
        thread::spawn(move || zip_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>) {
    let writer_opts = db::WriterOpts::default();
    let mut writer: db::Writer<Question> =
        db::Writer::new(NEW_DB_FILENAME, writer_opts).expect("new db writer");

    writer
        .load(&mut rx.iter())
        .unwrap_or_else(|e| panic!("db writer load, {e:#?}"));

    writer.finish().expect("db writer finish");

    println!("write done");
}