evict.rs

use std::{collections::HashMap, fs::File, path::Path};

use csv::StringRecord;
use itertools::Itertools;

fn open(path: impl AsRef<Path>) -> Vec<StringRecord> {
    let f = File::open(path).unwrap();
    let mut reader = csv::Reader::from_reader(f);
    let headers = reader.headers().cloned().unwrap();
    let records = reader.records().map(|res| res.unwrap()).collect_vec();
    println!("{:?}", headers);
    println!("records: {}", records.len());
    println!("example: {:?}", records[0]);
    records
}

fn main() {
    // time, actor id, sequence, table id, bytes
    let values = open("data/value.csv");
    // time, actor id, sequence, table id, count
    let counts = open("data/count.csv");

    // hash join
    let mut map = HashMap::new();
    for record in values {
        assert!(map
            .insert(
                (
                    record[0].to_owned(),
                    record[1].to_owned(),
                    record[2].to_owned(),
                    record[3].to_owned(),
                ),
                vec![record[4].to_owned()],
            )
            .is_none());
    }
    for record in counts {
        map.get_mut(&(
            record[0].to_owned(),
            record[1].to_owned(),
            record[2].to_owned(),
            record[3].to_owned(),
        ))
        .inspect(|v| assert_eq!(v.len(), 1))
        .unwrap()
        .push(record[4].to_owned());
    }

    // sort
    let mut records = map
        .into_iter()
        .map(|((time, actor, sequence, table), vc)| {
            (table, actor, sequence, time, vc[0].clone(), vc[1].clone())
        })
        .collect_vec();
    records.sort();

    // write
    let mut writer = csv::Writer::from_writer(File::create("data/output.csv").unwrap());
    writer
        .write_record(&["table", "actor", "sequence", "time", "bytes", "count"])
        .unwrap();
    for (table, actor, sequence, time, bytes, count) in records {
        writer
            .write_record(&[&table, &actor, &sequence, &time, &bytes, &count])
            .unwrap();
    }
    writer.flush().unwrap();
}