poetry_matcher.rs

#![feature(bool_to_option)]

use serde::Deserialize;
use std::path::PathBuf;
use regex::Regex;
use std::fs::{read_dir, File};
use opencc_rust::*;

#[derive(Debug, Deserialize)]
struct Poetry {
    author: String,
    paragraphs: Vec<String>,
    title: String
}

fn main() {
    let dir = PathBuf::from("../chinese-poetry/json");
    let file_pattern = Regex::new(r"poet\.\w*\.\d*\.json").unwrap();
    let find_pattern = Regex::new(r".*光.*量.*子.*").unwrap();

    let converter = OpenCC::new(DefaultConfig::T2S).unwrap();

    let files: Vec<_> = read_dir(dir).unwrap().into_iter().filter_map(|f|{
        let f = f.unwrap();
        file_pattern.is_match(f.file_name().to_str().unwrap()).then_some(f)
    }).collect();
    let j = files.len();
    for (i, file) in files.into_iter().enumerate() {
        if i % 10 == 0 {
            println!("Progress: {}", i as f32 / j as f32 * 100.0);
        }
        let f = File::open(file.path()).unwrap();
        let poets: Vec<Poetry> = serde_json::from_reader(f).unwrap();
        for poet in poets {
            for paragraph in &poet.paragraphs {
                if find_pattern.is_match(&*converter.convert(paragraph)) {
                    println!("{:#?}", poet);
                }
            }
        }
    }
}