Add amazon

This commit is contained in:
Rémi BERTHO 2020-06-21 10:36:13 +02:00
parent ed00ce7dec
commit f47b5f3b33
Signed by: dalan
GPG key ID: EE3B917931C07B64
3 changed files with 70 additions and 2 deletions

View file

@ -2,6 +2,7 @@ pub mod darty;
pub mod fnac; pub mod fnac;
pub mod du_bruit_dans_la_cuisine; pub mod du_bruit_dans_la_cuisine;
pub mod ldlc; pub mod ldlc;
pub mod amazon;
extern crate arraygen; extern crate arraygen;
extern crate url; extern crate url;
@ -22,6 +23,8 @@ pub trait PriceParser{
fn parse(&self, html : &Html) -> Result<PriceResult>; fn parse(&self, html : &Html) -> Result<PriceResult>;
} }
// @todo Macro générateur liste et tests
#[derive(Arraygen, Debug)] #[derive(Arraygen, Debug)]
#[gen_array(pub fn get_price: & dyn PriceParser)] #[gen_array(pub fn get_price: & dyn PriceParser)]
/// Represent the list of all the parser /// Represent the list of all the parser
@ -34,6 +37,8 @@ pub struct List {
du_bruit_dans_la_cuisine: du_bruit_dans_la_cuisine::DuBruitDansLaCuisine, du_bruit_dans_la_cuisine: du_bruit_dans_la_cuisine::DuBruitDansLaCuisine,
#[in_array(get_price)] #[in_array(get_price)]
ldlc: ldlc::LDLC, ldlc: ldlc::LDLC,
#[in_array(get_price)]
amazon: amazon::Amazon,
} }
impl List { impl List {
@ -43,7 +48,8 @@ impl List {
darty: darty::Darty::new()?, darty: darty::Darty::new()?,
fnac: fnac::Fnac::new()?, fnac: fnac::Fnac::new()?,
du_bruit_dans_la_cuisine: du_bruit_dans_la_cuisine::DuBruitDansLaCuisine::new()?, du_bruit_dans_la_cuisine: du_bruit_dans_la_cuisine::DuBruitDansLaCuisine::new()?,
ldlc: ldlc::LDLC::new()? ldlc: ldlc::LDLC::new()?,
amazon: amazon::Amazon::new()?
}) })
} }
} }
@ -51,5 +57,5 @@ impl List {
#[test] #[test]
fn test_parser_list() { fn test_parser_list() {
let parser_list = List::new().unwrap(); let parser_list = List::new().unwrap();
assert_eq!(parser_list.get_price().len(), 4); assert_eq!(parser_list.get_price().len(), 5);
} }

56
src/parser/amazon.rs Normal file
View file

@ -0,0 +1,56 @@
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 Amazon {
price_selector: Selector,
name_selector: Selector,
product_selector: Selector
}
impl PriceParser for Amazon {
fn new() -> Result<Self> {
Ok(Amazon {
price_selector: Selector::parse(r"#priceblock_ourprice").unwrap(),
name_selector: Selector::parse(r"#productTitle").unwrap(),
product_selector: Selector::parse(r".nav-search-label").unwrap()
})
}
fn can_parse(&self, url : &Url) -> bool {
url.host_str().unwrap_or("") == "www.amazon.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().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 = Amazon::new().unwrap();
assert!(fnac_parser.can_parse(&Url::parse("https://www.amazon.fr/AmazonBasics-Bo%C3%AEte-crayons-papier-pr%C3%A9taill%C3%A9s/dp/B071JM699B?ref_=ast_sto_dp").unwrap()));
assert!(fnac_parser.can_parse(&Url::parse("https://www.amazon.fr/AmazonBasics-Bo%C3%AEte-crayons-papier-pr%C3%A9taill%C3%A9s/dp/B071JM699B?ref_=ast_sto_dp").unwrap()));
assert!(fnac_parser.can_parse(&Url::parse("https://www.amazon.com").unwrap()) == false);
}

View file

@ -61,4 +61,10 @@ fn test_price_checker() {
let price_result = price_checker.get_price(Url::parse("https://www.ldlc.com/fiche/PB00335410.html").unwrap()).unwrap(); 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.name != "");
assert!(price_result.price != 0.); assert!(price_result.price != 0.);
// Amazon
let price_result = price_checker.get_price(Url::parse("https://www.amazon.fr/AmazonBasics-Bo%C3%AEte-crayons-papier-pr%C3%A9taill%C3%A9s/dp/B071JM699B?ref_=ast_sto_dp&th=1").unwrap()).unwrap();
assert!(price_result.name != "");
assert!(price_result.price != 0.);
assert!(price_result.product != "");
} }