dalan/price_checker
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Add foramt

Browse source
This commit is contained in:
Rémi BERTHO 2020-07-20 22:14:05 +02:00
parent 05c114cca4
commit 6233a754df
Signed by: dalan
GPG key ID: EE3B917931C07B64
11 changed files with 173 additions and 140 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
TODO.md~
View file

@ -1,14 +0,0 @@
# TODO
* [x] Récupération prix darty avec [scraper](https://crates.io/crates/scraper) et [reqwest](https://crates.io/crates/reqwest)
* [x] Ajout du support de la Fnac, …
* [x] Récupération URL ligne de commande avec [clap](https://crates.io/crates/clap)
* [ ] Ajout de SearchParser pour rechercher un article sur tous les parseurs
* [ ] Commande de liste des différents parseurs
* [ ] Ajout des pays avec [celes](https://crates.io/crates/celes) : recherche uniquement sur les parser du pays et parseur multi pays (amazon par exemple)
* [ ] Lecture des URLs depuis un fichier avec [toml](https://crates.io/crates/toml)
* [ ] Parallélisation des requêtes avce [rayon](rayon) ou reqwest asynchrone
* [ ] Écriture dans un fichier ODS avec [calamine](https://crates.io/crates/calamine)
* [ ] Notification si baisse de prix avec [notify-rust](https://crates.io/crates/notify-rust)
* [ ] Mail si baisse de prix avec [lettre](https://crates.io/crates/lettre)
* [ ] Lancement de la recherche régulière avec [Clokwerk](https://crates.io/crates/clokwerk)

6
src/.rustfmt.toml Normal file
View file

@ -0,0 +1,6 @@
max_width = 140
hard_tabs = true
tab_spaces = 4
newline_style = "Unix"
use_field_init_shorthand = true
edition = "2018"

48
src/main.rs
View file

@ -1,61 +1,58 @@
extern crate anyhow;
extern crate arraygen;
extern crate clap;
extern crate url;
extern crate arraygen;
pub mod price_result;
pub mod parser;
pub mod price_checker;
pub mod price_result;
use price_result::PriceResult;
use price_checker::PriceChecker;
use url::Url;
use anyhow::Result;
use clap::{Arg, App, SubCommand};
use clap::{App, Arg, SubCommand};
use price_checker::PriceChecker;
use price_result::PriceResult;
use url::Url;
fn main() -> Result<()> {
let matches = App::new("Price checker")
.version("0.1")
.author("Rémi BERTHO <remi.bertho@dalan.fr>")
.about("Check price")
.subcommand(SubCommand::with_name("check")
.subcommand(
SubCommand::with_name("check")
.about("Check from an URL")
.arg(Arg::with_name("URL")
.required(true)
.multiple(true)
.help("The URL to get price")))
.subcommand(SubCommand::with_name("search")
.arg(Arg::with_name("URL").required(true).multiple(true).help("The URL to get price")),
)
.subcommand(
SubCommand::with_name("search")
.about("Search an object")
.arg(Arg::with_name("name")
.required(true)
.multiple(true)
.help("The name of the object")))
.subcommand(SubCommand::with_name("list")
.about("List the parsers"))
.arg(Arg::with_name("name").required(true).multiple(true).help("The name of the object")),
)
.subcommand(SubCommand::with_name("list").about("List the parsers"))
.get_matches();
let price_checker = PriceChecker::new().unwrap();
match matches.subcommand() {
("check", Some(check_matches)) => {
for url_str in check_matches.values_of("URL").unwrap() {
let url = Url::parse(url_str)?;
println!("{}", price_checker.get_price(&url)?) ;
println!("{}", price_checker.get_price(&url)?);
}
}
},
("search", Some(check_matches)) => {
for name in check_matches.values_of("name").unwrap() {
let res = price_checker.search(name)? ;
let res = price_checker.search(name)?;
if res.is_empty() {
println!("«{}» not found", name);
} else {
for (parser_name, url) in res {
let price = price_checker.get_price(&url)?;
println!(" - {}", parser_name) ;
println!(" - {}", parser_name);
println!(" * {}", url);
println!(" * {}", price);
}
}
}
},
}
("list", _) => {
println!("Price parsers:");
for parser in price_checker.parser_list.get_price().iter() {
@ -65,11 +62,10 @@ fn main() -> Result<()> {
for parser in price_checker.parser_list.get_search().iter() {
println!(" - {}", parser.name());
}
},
}
_ => {
println!("{}", matches.usage());
},
}
}
Ok(())
}

40
src/parser.rs
View file

@ -1,35 +1,37 @@
pub mod darty;
pub mod fnac;
pub mod du_bruit_dans_la_cuisine;
pub mod ldlc;
pub mod amazon;
pub mod darty;
pub mod du_bruit_dans_la_cuisine;
pub mod fnac;
pub mod ldlc;
use scraper::Html;
use crate::price_result::PriceResult;
use arraygen::Arraygen;
use url::Url;
use anyhow::Result;
use arraygen::Arraygen;
use scraper::Html;
use url::Url;
pub trait Parser{
pub trait Parser {
/// Create the parser
fn new() -> Result<Self> where Self : Sized;
fn new() -> Result<Self>
where
Self: Sized;
/// Get the name
fn name(&self) -> &'static str;
}
/// Trait needed to get price from a specific website
pub trait PriceParser : Parser{
pub trait PriceParser: Parser {
/// Indicate if it can parse this URL
fn can_parse(&self, url : &Url) -> bool;
fn can_parse(&self, url: &Url) -> bool;
/// Parse the html into a price
fn parse_price(&self, html : &Html) -> Result<PriceResult>;
fn parse_price(&self, html: &Html) -> Result<PriceResult>;
}
pub trait SearchParser : Parser {
pub trait SearchParser: Parser {
/// Return the search URL
fn search_url(&self, name: &str) -> Url;
/// Return the first occurence of result of the page if any
fn search(&self, html : &Html) -> Result<Option<Url>>;
fn search(&self, html: &Html) -> Result<Option<Url>>;
}
macro_rules! gen_list {
@ -62,11 +64,11 @@ macro_rules! gen_list {
}
gen_list!(
[darty::Darty : get_price],
[fnac::Fnac : get_price],
[du_bruit_dans_la_cuisine::DuBruitDansLaCuisine : get_price],
[ldlc::LDLC : get_price],
[amazon::Amazon : get_price, get_search]
[darty::Darty: get_price],
[fnac::Fnac: get_price],
[du_bruit_dans_la_cuisine::DuBruitDansLaCuisine: get_price],
[ldlc::LDLC: get_price],
[amazon::Amazon: get_price, get_search]
);
#[test]

40
src/parser/amazon.rs
View file

@ -1,8 +1,8 @@
use super::{Parser, PriceParser, SearchParser};
use crate::PriceResult;
use scraper::{Selector, Html};
use anyhow::{anyhow, Result};
use scraper::{Html, Selector};
use url::Url;
use anyhow::{Result, anyhow};
#[derive(Debug)]
/// Parser for the darty website
@ -11,7 +11,7 @@ pub struct Amazon {
name_selector: Selector,
product_selector: Selector,
search_selector_1: Selector,
search_selector_2: Selector
search_selector_2: Selector,
}
impl Parser for Amazon {
@ -21,7 +21,7 @@ impl Parser for Amazon {
name_selector: Selector::parse(r"#productTitle").unwrap(),
product_selector: Selector::parse(r".nav-search-label").unwrap(),
search_selector_1: Selector::parse(r".rush-component[data-component-type=s-product-image]").unwrap(),
search_selector_2: Selector::parse(r".a-link-normal").unwrap()
search_selector_2: Selector::parse(r".a-link-normal").unwrap(),
})
}
@ -31,16 +31,21 @@ impl Parser for Amazon {
}
impl PriceParser for Amazon {
fn can_parse(&self, url : &Url) -> bool {
fn can_parse(&self, url: &Url) -> bool {
url.host_str().unwrap_or("") == "www.amazon.fr"
}
fn parse_price(&self, html : &Html) -> Result<PriceResult> {
fn parse_price(&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()?;
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"))?;
@ -53,19 +58,22 @@ impl PriceParser for Amazon {
Ok(PriceResult {
name: name.to_owned(),
product: family.to_owned(),
price
price,
})
}
}
impl SearchParser for Amazon {
fn search_url(&self, name: &str) -> Url {
Url::parse(& format!("https://www.amazon.fr/s?k={}", name)).unwrap()
Url::parse(&format!("https://www.amazon.fr/s?k={}", name)).unwrap()
}
fn search(&self, html : &Html) -> Result<Option<Url>> {
fn search(&self, html: &Html) -> Result<Option<Url>> {
if let Some(search_element_1) = html.select(&self.search_selector_1).next() {
let search_element_2 = search_element_1.select(&self.search_selector_2).next().ok_or(anyhow!("No search element 2"))?;
let search_element_2 = search_element_1
.select(&self.search_selector_2)
.next()
.ok_or(anyhow!("No search element 2"))?;
let path_url = search_element_2.value().attr("href").ok_or(anyhow!("No link element"))?;
let mut url = Url::parse("https://www.amazon.fr").unwrap();
url.set_path(path_url.split("/ref").next().unwrap_or(path_url));
@ -79,7 +87,11 @@ impl SearchParser for Amazon {
#[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").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").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").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").unwrap()
));
assert!(fnac_parser.can_parse(&Url::parse("https://www.amazon.com").unwrap()) == false);
}

18
src/parser/darty.rs
View file

@ -1,15 +1,15 @@
use super::{Parser, PriceParser};
use crate::PriceResult;
use scraper::{Selector, Html};
use anyhow::{anyhow, Result};
use scraper::{Html, Selector};
use url::Url;
use anyhow::{Result, anyhow};
#[derive(Debug)]
/// Parser for the darty website
pub struct Darty {
price_selector: Selector,
name_selector: Selector,
product_selector: Selector
product_selector: Selector,
}
impl Parser for Darty {
@ -17,7 +17,7 @@ impl Parser for Darty {
Ok(Darty {
price_selector: Selector::parse(r#".darty_prix"#).unwrap(),
name_selector: Selector::parse(r#".product_name"#).unwrap(),
product_selector: Selector::parse(r#".product_family"#).unwrap()
product_selector: Selector::parse(r#".product_family"#).unwrap(),
})
}
@ -27,16 +27,16 @@ impl Parser for Darty {
}
impl PriceParser for Darty {
fn can_parse(&self, url : &Url) -> bool {
fn can_parse(&self, url: &Url) -> bool {
url.host_str().unwrap_or("") == "www.darty.com"
}
fn parse_price(&self, html : &Html) -> Result<PriceResult> {
fn parse_price(&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_end_matches('€').parse()?;
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_end_matches('€').parse()?;
let price = price_ent as f64 + (price_dec as f64) / 100.;
// Get name
@ -50,7 +50,7 @@ impl PriceParser for Darty {
Ok(PriceResult {
name: name.to_owned(),
product: family.to_owned(),
price
price,
})
}
}

20
src/parser/du_bruit_dans_la_cuisine.rs
View file

@ -1,8 +1,8 @@
use super::{PriceParser, Parser};
use super::{Parser, PriceParser};
use crate::PriceResult;
use scraper::{Selector, Html};
use anyhow::{anyhow, Result};
use scraper::{Html, Selector};
use url::Url;
use anyhow::{Result, anyhow};
#[derive(Debug)]
/// Parser for the darty website
@ -25,15 +25,21 @@ impl Parser for DuBruitDansLaCuisine {
}
impl PriceParser for DuBruitDansLaCuisine {
fn can_parse(&self, url : &Url) -> bool {
fn can_parse(&self, url: &Url) -> bool {
url.host_str().unwrap_or("") == "www.dubruitdanslacuisine.fr"
}
fn parse_price(&self, html : &Html) -> Result<PriceResult> {
fn parse_price(&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()?;
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"))?;
@ -42,7 +48,7 @@ impl PriceParser for DuBruitDansLaCuisine {
Ok(PriceResult {
name: name.to_owned(),
product: "Cuisine".to_owned(),
price
price,
})
}
}

24
src/parser/fnac.rs
View file

@ -1,23 +1,23 @@
use super::{Parser, PriceParser};
use crate::PriceResult;
use scraper::{Selector, Html};
use anyhow::{anyhow, Result};
use scraper::{Html, Selector};
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
product_selector: Selector,
}
impl Parser for Fnac{
impl Parser 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()
product_selector: Selector::parse(r#".f-productHeader-subTitleLink"#).unwrap(),
})
}
@ -27,16 +27,16 @@ impl Parser for Fnac{
}
impl PriceParser for Fnac {
fn can_parse(&self, url : &Url) -> bool {
fn can_parse(&self, url: &Url) -> bool {
url.host_str().unwrap_or("") == "www.fnac.com"
}
fn parse_price(&self, html : &Html) -> Result<PriceResult> {
fn parse_price(&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_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
@ -50,7 +50,7 @@ impl PriceParser for Fnac {
Ok(PriceResult {
name: name.to_owned(),
product: family.to_owned(),
price
price,
})
}
}
@ -58,7 +58,9 @@ impl PriceParser for Fnac {
#[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("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);
}

16
src/parser/ldlc.rs
View file

@ -1,8 +1,8 @@
use super::{PriceParser, Parser};
use super::{Parser, PriceParser};
use crate::PriceResult;
use scraper::{Selector, Html};
use anyhow::{anyhow, Result};
use scraper::{Html, Selector};
use url::Url;
use anyhow::{Result, anyhow};
#[derive(Debug)]
/// Parser for the darty website
@ -25,16 +25,16 @@ impl Parser for LDLC {
}
impl PriceParser for LDLC {
fn can_parse(&self, url : &Url) -> bool {
fn can_parse(&self, url: &Url) -> bool {
url.host_str().unwrap_or("") == "www.ldlc.com"
}
fn parse_price(&self, html : &Html) -> Result<PriceResult> {
fn parse_price(&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_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
@ -44,7 +44,7 @@ impl PriceParser for LDLC {
Ok(PriceResult {
name: name.to_owned(),
product: "High-tech".to_owned(),
price
price,
})
}
}

47
src/price_checker.rs
View file

@ -1,19 +1,19 @@
extern crate reqwest;
extern crate scraper;
use reqwest::blocking::Client;
use scraper::Html;
use crate::parser;
use crate::price_result::PriceResult;
use anyhow::{anyhow, Result};
use reqwest::blocking::Client;
use scraper::Html;
use url::Url;
use anyhow::{Result, anyhow};
const USER_AGENT: &str = "Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Firefox/78.0";
/// Struct used to get price from a website
pub struct PriceChecker {
client: Client,
pub parser_list: parser::List
pub parser_list: parser::List,
}
impl PriceChecker {
@ -22,19 +22,24 @@ impl PriceChecker {
let client = reqwest::blocking::Client::builder().user_agent(USER_AGENT).build()?;
Ok(PriceChecker {
client,
parser_list: parser::List::new()?
parser_list: parser::List::new()?,
})
}
fn get_html(&self, url : &Url) -> Result<Html> {
fn get_html(&self, url: &Url) -> Result<Html> {
let response = self.client.get(url.clone()).send()?;
let text = response.text()?;
Ok(Html::parse_document(&text))
}
/// Get a price from an URL
pub fn get_price(&self, url : &Url) -> Result<PriceResult> {
let parser = *self.parser_list.get_price().iter().find(|p| p.can_parse(&url)).ok_or(anyhow!("No parser can parse {}", url))?;
pub fn get_price(&self, url: &Url) -> Result<PriceResult> {
let parser = *self
.parser_list
.get_price()
.iter()
.find(|p| p.can_parse(&url))
.ok_or(anyhow!("No parser can parse {}", url))?;
let document = self.get_html(url)?;
Ok(parser.parse_price(&document)?)
}
@ -64,23 +69,41 @@ fn test_price_checker() {
assert!(price_result.product != "");
// Test fnac
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();
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();
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();
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.);
// 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();
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 != "");

2
src/price_result.rs
View file

@ -8,7 +8,7 @@ pub struct PriceResult {
/// The product type
pub product: String,
/// The price
pub price: f64
pub price: f64,
}
impl fmt::Display for PriceResult {