chgk_ledb/app/src/main.rs

181 lines
4.9 KiB
Rust
Raw Normal View History

2022-09-17 18:08:13 +00:00
use clap::{Parser, Subcommand};
2022-09-17 15:45:41 +00:00
use rand::seq::IteratorRandom;
2022-10-05 13:21:52 +00:00
use std::io;
2022-09-17 15:45:41 +00:00
use std::time::Instant;
2022-10-07 23:02:27 +00:00
use std::{fs, sync::mpsc, thread};
2022-11-10 18:43:15 +00:00
use chgk_ledb_lib::db;
use chgk_ledb_lib::questions;
use chgk_ledb_lib::source;
2022-10-05 11:55:05 +00:00
use crate::questions::{Question, QuestionsConverter};
use crate::source::ReadSourceQuestionsBatches;
2023-08-25 12:28:19 +00:00
use chgk_ledb_lib::util::ErrorToString;
2022-10-05 11:55:05 +00:00
2022-09-18 09:08:25 +00:00
const ZIP_FILENAME: &str = "json.zip";
2023-01-03 20:51:09 +00:00
const NEW_DB_FILENAME: &str = "db.dat";
2022-09-18 09:08:25 +00:00
2022-09-17 18:08:13 +00:00
#[derive(Subcommand, Debug)]
enum Command {
Write,
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,
},
}
#[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,
}
2023-08-25 12:28:19 +00:00
fn main() {
let args = Cli::parse();
2022-10-10 12:44:32 +00:00
2023-08-25 12:28:19 +00:00
let mut action: Box<dyn FnOnce()> = match &args.command {
Command::Write => Box::new(write_db),
Command::Print { id } => {
let get_question = Box::new(|| read_from_db(*id));
Box::new(|| print_question_from(get_question))
2022-10-07 12:43:22 +00:00
}
2023-08-25 12:28:19 +00:00
Command::ZipPrint { file_num, num } => {
let get_question = Box::new(|| read_from_zip(*file_num, *num));
Box::new(|| print_question_from(get_question))
}
};
if args.measure {
action = Box::new(|| measure_and_print(action));
2022-10-07 12:43:22 +00:00
}
2023-08-25 12:28:19 +00:00
action();
}
// 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);
2022-10-05 13:21:52 +00:00
}
2022-09-17 15:45:41 +00:00
2022-09-17 18:08:13 +00:00
fn print_question_from<F>(get_q: F)
where
2023-08-25 12:28:19 +00:00
F: FnOnce() -> Result<Question, String>,
2022-09-17 18:08:13 +00:00
{
2023-03-29 10:00:17 +00:00
let q = get_q().expect("question not found");
2022-09-17 18:08:13 +00:00
println!("{:#?}", q)
2022-09-17 15:45:41 +00:00
}
2023-08-25 12:28:19 +00:00
fn read_from_zip(file_num: usize, mut num: usize) -> Result<Question, String> {
2022-10-04 19:58:57 +00:00
let mut rng = rand::thread_rng();
2023-08-25 12:28:19 +00:00
let zip_file = fs::File::open(ZIP_FILENAME).str_err()?;
2022-09-17 15:45:41 +00:00
let zip_reader = io::BufReader::new(zip_file);
2023-08-25 12:28:19 +00:00
let archive = zip::ZipArchive::new(zip_reader).str_err()?;
2022-09-17 15:45:41 +00:00
2022-10-04 19:58:57 +00:00
let mut source_questions = archive.source_questions();
2022-10-05 11:55:05 +00:00
let (filename, batch) = if file_num == 0 {
2023-08-25 12:28:19 +00:00
source_questions
.choose(&mut rng)
.ok_or("rand choose".to_string())?
2022-10-04 19:58:57 +00:00
} else {
2023-08-25 12:28:19 +00:00
source_questions
.nth(file_num - 1)
.ok_or(format!("file nth #{file_num} => None"))?
2022-10-04 19:58:57 +00:00
};
2023-08-25 12:28:19 +00:00
let mut batch = batch.map_err(|e| format!("get batch from file #{file_num} => {e}"))?;
2022-10-05 11:55:05 +00:00
batch.filename = filename;
2022-10-04 19:58:57 +00:00
let questions: Vec<Question> = batch.into();
2022-09-17 18:08:13 +00:00
if num == 0 {
2023-08-25 12:28:19 +00:00
num = (1..=questions.len())
.choose(&mut rng)
.ok_or("rand choose".to_string())?;
2022-09-17 18:08:13 +00:00
}
2023-08-25 12:28:19 +00:00
Ok(questions[num - 1].clone())
2022-09-17 15:45:41 +00:00
}
2023-08-25 12:28:19 +00:00
fn read_from_db(id: u32) -> Result<Question, String> {
let reader: db::Reader<Question> = db::Reader::new(NEW_DB_FILENAME, 2048)?;
2022-10-10 12:44:32 +00:00
2023-08-25 12:28:19 +00:00
let len = reader.len();
2023-03-29 10:00:17 +00:00
let mut questions = reader.into_iter();
2022-10-06 18:22:19 +00:00
2023-08-25 12:28:19 +00:00
let question = match id {
2022-10-07 08:53:17 +00:00
0 => {
let mut rng = rand::thread_rng();
2023-08-25 12:28:19 +00:00
questions
.choose(&mut rng)
.ok_or(format!("rand choose, len = {len}"))?
2022-10-07 08:53:17 +00:00
}
2023-08-25 12:28:19 +00:00
_ => questions
.nth((id - 1) as usize)
.ok_or(format!("get nth #{id} => None"))?,
};
Ok(question)
2022-10-06 18:22:19 +00:00
}
2023-08-25 12:28:19 +00:00
2023-03-29 10:00:17 +00:00
fn write_db() {
2022-10-06 18:22:19 +00:00
let (tx, rx) = mpsc::channel::<Question>();
[
2022-10-10 13:13:27 +00:00
thread::spawn(move || zip_reader_task(tx)),
2023-03-29 10:00:17 +00:00
thread::spawn(move || db_writer_task(rx)),
2022-10-06 18:22:19 +00:00
]
.into_iter()
.for_each(|handle| handle.join().expect("thread panic"));
println!("all done");
}
2023-08-25 12:28:19 +00:00
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");
}
2023-03-29 10:00:17 +00:00
fn db_writer_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");
2022-10-09 10:38:25 +00:00
2022-10-09 15:26:00 +00:00
writer
2022-10-10 12:44:32 +00:00
.load(&mut rx.iter())
2022-10-09 21:08:57 +00:00
.unwrap_or_else(|e| panic!("db writer load, {e:#?}"));
2022-10-09 15:26:00 +00:00
writer.finish().expect("db writer finish");
2022-10-06 18:22:19 +00:00
println!("write done");
}