price_checker/src/parser/fnac.rs

58 lines
2.0 KiB
Rust

use super::PriceParser;
use crate::PriceResult;
use scraper::{Selector, Html};
use url::Url;
use anyhow::{Result, anyhow};
#[derive(Debug)]
/// Parser for the darty website
pub struct Fnac {
price_selector: Selector,
name_selector: Selector,
product_selector: Selector
}
impl PriceParser for Fnac {
fn new() -> Result<Self> {
Ok(Fnac {
price_selector: Selector::parse(r#".f-priceBox-price.checked"#).unwrap(),
name_selector: Selector::parse(r#".f-productHeader-Title"#).unwrap(),
product_selector: Selector::parse(r#".f-productHeader-subTitleLink"#).unwrap()
})
}
fn can_parse(&self, url : &Url) -> bool {
url.host_str().unwrap_or("") == "www.fnac.com"
}
fn parse(&self, html : &Html) -> Result<PriceResult> {
// Get price
let price_element = html.select(&self.price_selector).next().ok_or(anyhow!("No price element"))?;
let mut price_text_it = price_element.text();
let price_ent : u32 = price_text_it.next().unwrap_or("0").trim_end_matches(',').parse()?;
let price_dec : u32 = price_text_it.next().unwrap_or("0").trim_start_matches('€').parse()?;
let price = price_ent as f64 + (price_dec as f64) / 100.;
// Get name
let name_element = html.select(&self.name_selector).next().ok_or(anyhow!("No name element"))?;
let name = name_element.text().next().unwrap_or("").trim().replace('\n', "-");
// Get product
let family_element = html.select(&self.product_selector).next().ok_or(anyhow!("No product element"))?;
let family = family_element.text().next().unwrap_or("").trim().replace('\n', "-");
Ok(PriceResult {
name: name.to_owned(),
product: family.to_owned(),
price
})
}
}
#[test]
fn test_parser_fnac() {
let fnac_parser = Fnac::new().unwrap();
assert!(fnac_parser.can_parse(&Url::parse("https://www.fnac.com/Apple-iPhone-XS-64-Go-5-8-Argent/a12849718/w-4?CtoPid=488371").unwrap()));
assert!(fnac_parser.can_parse(&Url::parse("http://www.fnac.com/Apple-iPhone-XS-64-Go-5-8-Argent/a12849718/w-4?CtoPid=488371").unwrap()));
assert!(fnac_parser.can_parse(&Url::parse("https://www.fnace.com").unwrap()) == false);
}