Add LDLC and Du Bruit Dans la Cuisine

This commit is contained in:
Rémi BERTHO 2020-06-15 17:54:45 +02:00
parent fcf0cd2d96
commit ed00ce7dec
Signed by: dalan
GPG key ID: EE3B917931C07B64
4 changed files with 123 additions and 3 deletions

View file

@ -1,5 +1,7 @@
pub mod darty;
pub mod fnac;
pub mod du_bruit_dans_la_cuisine;
pub mod ldlc;
extern crate arraygen;
extern crate url;
@ -28,6 +30,10 @@ pub struct List {
darty: darty::Darty,
#[in_array(get_price)]
fnac: fnac::Fnac,
#[in_array(get_price)]
du_bruit_dans_la_cuisine: du_bruit_dans_la_cuisine::DuBruitDansLaCuisine,
#[in_array(get_price)]
ldlc: ldlc::LDLC,
}
impl List {
@ -35,7 +41,9 @@ impl List {
pub fn new() -> Result<Self> {
Ok(List {
darty: darty::Darty::new()?,
fnac: fnac::Fnac::new()?
fnac: fnac::Fnac::new()?,
du_bruit_dans_la_cuisine: du_bruit_dans_la_cuisine::DuBruitDansLaCuisine::new()?,
ldlc: ldlc::LDLC::new()?
})
}
}
@ -43,5 +51,5 @@ impl List {
#[test]
fn test_parser_list() {
let parser_list = List::new().unwrap();
assert_eq!(parser_list.get_price().len(), 2);
assert_eq!(parser_list.get_price().len(), 4);
}

View file

@ -0,0 +1,50 @@
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 DuBruitDansLaCuisine {
price_selector: Selector,
name_selector: Selector,
}
impl PriceParser for DuBruitDansLaCuisine {
fn new() -> Result<Self> {
Ok(DuBruitDansLaCuisine {
price_selector: Selector::parse(r#".price"#).unwrap(),
name_selector: Selector::parse(r#".product.item"#).unwrap(),
})
}
fn can_parse(&self, url : &Url) -> bool {
url.host_str().unwrap_or("") == "www.dubruitdanslacuisine.fr"
}
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 : f64 = price_text_it.next().unwrap_or("0.").trim_end_matches("").trim().replace(',', ".").parse()?;
// Get name
let name_element = html.select(&self.name_selector).next().ok_or(anyhow!("No name element"))?;
let name = name_element.text().nth(1).unwrap_or("").trim();
Ok(PriceResult {
name: name.to_owned(),
product: "Cuisine".to_owned(),
price
})
}
}
#[test]
fn test_parser_du_bruit_dans_la_cuisine() {
let parser = DuBruitDansLaCuisine::new().unwrap();
assert!(parser.can_parse(&Url::parse("https://www.dubruitdanslacuisine.fr/tapis-a-patisserie-40-62-14377-p").unwrap()));
assert!(parser.can_parse(&Url::parse("https://www.dubruitdanslacuisine.fr/tapis-a-patisserie-40-62-14377-p").unwrap()));
assert!(parser.can_parse(&Url::parse("https://www.dubrutdanslacuisine.fr/").unwrap()) == false);
}

52
src/parser/ldlc.rs Normal file
View file

@ -0,0 +1,52 @@
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 LDLC {
price_selector: Selector,
name_selector: Selector,
}
impl PriceParser for LDLC {
fn new() -> Result<Self> {
Ok(LDLC {
price_selector: Selector::parse(r#".price"#).unwrap(),
name_selector: Selector::parse(r#".title-1"#).unwrap(),
})
}
fn can_parse(&self, url : &Url) -> bool {
url.host_str().unwrap_or("") == "www.ldlc.com"
}
fn parse(&self, html : &Html) -> Result<PriceResult> {
// Get price
let price_element = html.select(&self.price_selector).nth(4).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").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();
Ok(PriceResult {
name: name.to_owned(),
product: "High-tech".to_owned(),
price
})
}
}
#[test]
fn test_parser_du_bruit_dans_la_cuisine() {
let parser = LDLC::new().unwrap();
assert!(parser.can_parse(&Url::parse("https://www.ldlc.com/fiche/PB00335410.html").unwrap()));
assert!(parser.can_parse(&Url::parse("http://www.ldlc.com/fiche/PB00335410.html").unwrap()));
assert!(parser.can_parse(&Url::parse("https://www.ldlv.com").unwrap()) == false);
}

View file

@ -47,8 +47,18 @@ fn test_price_checker() {
assert!(price_result.product != "");
// Test fnac
let price_result = price_checker.get_price(Url::parse("https://www.fnac.com/PC-Gaming-Milenium-MM1-Ekko-R207-A536X-W-S-R48-B-D1-AMD-Ryzen-5-32-Go-RAM-500-Go-SSD-1-To-SATA-Noir/a14620943/w-4").unwrap()).unwrap();
let price_result = price_checker.get_price(Url::parse("https://www.fnac.com/a12584732/Kaamelott-Les-Six-Livres-L-integrale-de-la-serie-Coffret-Blu-ray-Alexandre-Astier-Blu-ray").unwrap()).unwrap();
assert!(price_result.name != "");
assert!(price_result.price != 0.);
assert!(price_result.product != "");
// Test du bruis dans la cuisine
let price_result = price_checker.get_price(Url::parse("https://www.dubruitdanslacuisine.fr/tapis-a-patisserie-40-62-14377-p").unwrap()).unwrap();
assert!(price_result.name != "");
assert!(price_result.price != 0.);
// LDLC
let price_result = price_checker.get_price(Url::parse("https://www.ldlc.com/fiche/PB00335410.html").unwrap()).unwrap();
assert!(price_result.name != "");
assert!(price_result.price != 0.);
}