scraper-rs: dia

scraper-rs/src/main.rs

@@ -1,6 +1,7 @@
 use again::RetryPolicy;
 use async_channel::{Receiver, Sender};
 use nanoid::nanoid;
+use reqwest::Url;
 use rusqlite::Connection;
 use simple_error::{bail, SimpleError};
 use std::{
@@ -88,6 +89,7 @@ async fn fetch_and_parse(
     client: &reqwest::Client,
     url: String,
 ) -> Result<PrecioPoint, anyhow::Error> {
+    let url_p = Url::parse(&url).unwrap();
     let policy = RetryPolicy::exponential(Duration::from_millis(300))
         .with_max_retries(10)
         .with_jitter(true);
@@ -106,7 +108,11 @@ async fn fetch_and_parse(
 
     let maybe_point = {
         let dom = tl::parse(&body, tl::ParserOptions::default()).map_err(FetchError::Tl)?;
-        sites::carrefour::parse(url, &dom)
+        match url_p.host_str().unwrap() {
+            "www.carrefour.com.ar" => sites::carrefour::parse(url, &dom),
+            "diaonline.supermercadosdia.com.ar" => sites::dia::parse(url, &dom),
+            s => bail!("Unknown host {}", s),
+        }
     };
 
     let point = match maybe_point {

scraper-rs/src/sites/carrefour.rs

@@ -8,9 +8,7 @@ use crate::PrecioPoint;
 use super::vtex::find_product_ld;
 
 pub fn parse(url: String, dom: &tl::VDom) -> Result<PrecioPoint, anyhow::Error> {
-    let precio_centavos = common::get_meta_content(dom, "product:price:amount")
-        .map(|s| s.parse::<f64>().map(|f| (f * 100.0) as u64))
-        .transpose()?;
+    let precio_centavos = common::price_from_meta(dom)?;
 
     let in_stock = match common::get_meta_content(dom, "product:availability") {
         Some(s) => match s.as_ref() {

scraper-rs/src/sites/common.rs

@@ -10,3 +10,10 @@ pub fn get_meta_content<'a>(dom: &'a VDom<'a>, prop: &str) -> Option<Cow<'a, str
         .and_then(|tag| tag.attributes().get("content").flatten())
         .map(|s| s.as_utf8_str())
 }
+
+pub fn price_from_meta(dom: &tl::VDom<'_>) -> Result<Option<u64>, anyhow::Error> {
+    let precio_centavos = get_meta_content(dom, "product:price:amount")
+        .map(|s| s.parse::<f64>().map(|f| (f * 100.0) as u64))
+        .transpose()?;
+    Ok(precio_centavos)
+}

scraper-rs/src/sites/dia.rs

@@ -0,0 +1,43 @@
+use anyhow::Context;
+use simple_error::bail;
+use simple_error::SimpleError;
+
+use crate::sites::common;
+use crate::sites::vtex;
+use crate::PrecioPoint;
+
+use super::vtex::find_product_ld;
+use super::vtex::AvailabilityLd;
+
+pub fn parse(url: String, dom: &tl::VDom) -> Result<PrecioPoint, anyhow::Error> {
+    let ean = common::get_meta_content(dom, "product:retailer_item_id")
+        .context("Parsing EAN")?
+        .to_string();
+    let precio_centavos = common::price_from_meta(dom)?;
+
+    let (name, image_url, in_stock) = match find_product_ld(dom) {
+        Some(pm) => {
+            let p = pm?;
+            (
+                Some(p.name),
+                Some(p.image),
+                Some(
+                    p.offers.offers.first().context("No offer")?.availability
+                        == AvailabilityLd::InStock,
+                ),
+            )
+        }
+        None => bail!("No JSON/LD"),
+    };
+
+    Ok(PrecioPoint {
+        ean,
+        fetched_at: crate::now_sec(),
+        in_stock,
+        name,
+        image_url,
+        parser_version: 5,
+        precio_centavos,
+        url,
+    })
+}

scraper-rs/src/sites/mod.rs

@@ -1,3 +1,4 @@
 pub mod carrefour;
 mod common;
+pub mod dia;
 mod vtex;

scraper-rs/src/sites/vtex.rs

@@ -79,7 +79,7 @@ pub struct OfferLd {
 pub enum OfferTypeLd {
     Offer,
 }
-#[derive(Deserialize)]
+#[derive(Deserialize, PartialEq)]
 pub enum AvailabilityLd {
     #[serde(rename = "http://schema.org/InStock")]
     InStock,