네이버 한자사전 고사성어 크롤러

Cargo.toml

[package]
name = "hanja-idiom"
version = "0.1.0"
authors = ["kiwiyou <[email protected]>"]
edition = "2018"

[dependencies]
reqwest = "0.9"
unhtml = "0.7"
unhtml_derive = "0.7"
serde = "1.0"
serde_derive = "1.0"
csv = "1.1"
rayon = "1.3"

main.rs

use rayon::prelude::*;
use reqwest::get;
use serde_derive::*;
use unhtml::{self, FromHtml};
use unhtml_derive::*;

#[derive(FromHtml)]
#[html(selector = ".tab_submenu")]
struct OnsetTab {
    #[html(selector = "li")]
    onsets: Vec<Onset>,
}

#[derive(FromHtml)]
struct Onset;

#[derive(FromHtml)]
#[html(selector = ".sub_word")]
struct InitialTab {
    #[html(selector = "span", attr = "inner")]
    initials: Vec<String>,
}

#[derive(Debug, FromHtml)]
#[html(selector = "#content")]
struct WordList {
    #[html(selector = "dd:not([class])", attr = "inner")]
    readings: Vec<String>,
    #[html(selector = "dd.meaning", attr = "inner")]
    meanings: Vec<String>,
    #[html(selector = ".paginate > *:not([class])", attr = "inner")]
    pages: Vec<usize>,
    #[html(selector = ".paginate > .next", attr = "inner")]
    next: Option<String>,
}

fn parse_onset_tab() -> OnsetTab {
    let html = get("https://hanja.dict.naver.com/category/idiom")
        .unwrap()
        .text()
        .unwrap();
    let tab = OnsetTab::from_html(&html).unwrap();
    tab
}

fn parse_initial_tab(index: usize) -> InitialTab {
    let query = format!("https://hanja.dict.naver.com/category/idiom?idx={}", index);
    let html = get(&query).unwrap().text().unwrap();
    let tab = InitialTab::from_html(&html).unwrap();
    tab
}

fn parse_word_list<S: AsRef<str>>(initial: S, page: usize) -> WordList {
    let query = format!(
        "https://hanja.dict.naver.com/category/idiom?q={}&pageNo={}",
        initial.as_ref(),
        page
    );
    let html = get(&query).unwrap().text().unwrap();
    let tab = WordList::from_html(&html).unwrap();
    tab
}

#[derive(Debug, Clone, Serialize)]
#[serde(rename_all = "PascalCase")]
struct Idiom {
    reading: String,
    meaning: String,
}

fn crawl_for_initial<S: AsRef<str>>(initial: S) -> Vec<Idiom> {
    let initial_ref = initial.as_ref();
    let sample_list = sample_for_initial(initial_ref);
    sample_list
        .par_iter()
        .flat_map(|sample| {
            let first = sample
                .readings
                .par_iter()
                .zip(sample.meanings.par_iter())
                .map(|(reading, meaning)| Idiom {
                    reading: reading.clone(),
                    meaning: meaning.clone(),
                })
                .collect::<Vec<_>>();
            let from_second = sample
                .pages
                .par_iter()
                .skip(1)
                .map(|page| parse_word_list(initial_ref, *page))
                .flat_map(|list| {
                    list.readings
                        .par_iter()
                        .zip(list.meanings.par_iter())
                        .map(|(reading, meaning)| Idiom {
                            reading: reading.clone(),
                            meaning: meaning.clone(),
                        })
                        .collect::<Vec<_>>()
                });
            first
                .par_iter()
                .cloned()
                .chain(from_second)
                .collect::<Vec<_>>()
        })
        .collect()
}

fn sample_for_initial<S: AsRef<str>>(initial: S) -> Vec<WordList> {
    let initial_ref = initial.as_ref();
    let mut sample_page = 1;
    let mut sample_list = Vec::new();
    loop {
        let sample = parse_word_list(initial_ref, sample_page);
        let is_none = sample.next.is_none();
        sample_list.push(sample);
        if is_none {
            break;
        }
        sample_page += 10;
    }
    sample_list
}

fn main() {
    println!("[+] crawling onset list...");
    let onset_tab = parse_onset_tab();
    let initial_tabs = onset_tab
        .onsets
        .par_iter()
        .enumerate()
        .map(|(i, _)| parse_initial_tab(i));
    println!("[+] crawling idiom list...");
    let mut idioms = initial_tabs
        .flat_map(|tab| {
            tab.initials
                .par_iter()
                .flat_map(|initial| crawl_for_initial(initial))
                .collect::<Vec<_>>()
        })
        .collect::<Vec<_>>();
    idioms.sort_unstable_by(|a, b| a.reading.partial_cmp(&b.reading).unwrap());
    println!("[+] saving idioms...");
    let mut writer = csv::WriterBuilder::new()
        .delimiter(b',')
        .quote_style(csv::QuoteStyle::NonNumeric)
        .from_path("idiom.csv")
        .unwrap();
    for idiom in idioms.iter() {
        writer.serialize(idiom).unwrap();
    }
    writer.flush().unwrap();
    println!("[+] list saved into idiom.csv!");
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn onset_tab_should_parse_correct_onset_list() {
        let tab = parse_onset_tab();
        assert_eq!(14, tab.onsets.len());
    }

    #[test]
    fn initial_tab_should_parse_correct_initial_list() {
        let initial_tab = parse_initial_tab(0);
        assert_eq!(55, initial_tab.initials.len());
    }

    #[test]
    fn word_list_should_have_equal_lengths_for_readings_and_meanings() {
        let word_list = parse_word_list("가", 0);
        assert_eq!(word_list.readings.len(), word_list.meanings.len());
    }
}