From 1118bcf75d4f596392d466c0a69f73b95eaaab5f Mon Sep 17 00:00:00 2001
From: Nulo <git@nulo.in>
Date: Tue, 6 Feb 2024 23:20:04 -0300
Subject: [PATCH] bringup farmacity

---
 db-datos/supermercado.ts                |  6 +++
 scraper-rs/src/main.rs                  |  4 ++
 scraper-rs/src/sites/farmacity.rs       | 50 +++++++++++++++++++++++++
 scraper-rs/src/sites/mod.rs             |  1 +
 scraper-rs/src/supermercado.rs          |  2 +
 sitio/src/routes/ean/[ean]/+page.svelte |  1 +
 6 files changed, 64 insertions(+)
 create mode 100644 scraper-rs/src/sites/farmacity.rs

diff --git a/db-datos/supermercado.ts b/db-datos/supermercado.ts
index ec4c8c3..992e1bf 100644
--- a/db-datos/supermercado.ts
+++ b/db-datos/supermercado.ts
@@ -3,25 +3,31 @@ export enum Supermercado {
   Carrefour = "Carrefour",
   Coto = "Coto",
   Jumbo = "Jumbo",
+  Farmacity = "Farmacity",
 }
 export const supermercados: Supermercado[] = [
   Supermercado.Carrefour,
   Supermercado.Coto,
   Supermercado.Dia,
   Supermercado.Jumbo,
+  Supermercado.Farmacity,
 ];
 export const hosts: { [host: string]: Supermercado } = {
   "diaonline.supermercadosdia.com.ar": Supermercado.Dia,
   "www.carrefour.com.ar": Supermercado.Carrefour,
   "www.cotodigital3.com.ar": Supermercado.Coto,
   "www.jumbo.com.ar": Supermercado.Jumbo,
+  "www.farmacity.com": Supermercado.Farmacity,
 };
 export const hostBySupermercado = Object.fromEntries(
   Object.entries(hosts).map(([a, b]) => [b, a])
 ) as Record<Supermercado, string>;
+
+// también actualizar en sitio/src/routes/ean/[ean]/+page.svelte
 export const colorBySupermercado: { [supermercado in Supermercado]: string } = {
   [Supermercado.Dia]: "#d52b1e",
   [Supermercado.Carrefour]: "#19549d",
   [Supermercado.Coto]: "#e20025",
   [Supermercado.Jumbo]: "#2dc850",
+  [Supermercado.Farmacity]: "#EF7603",
 };
diff --git a/scraper-rs/src/main.rs b/scraper-rs/src/main.rs
index b5d75b9..0f9efbf 100644
--- a/scraper-rs/src/main.rs
+++ b/scraper-rs/src/main.rs
@@ -285,6 +285,7 @@ async fn get_urls(supermercado: &Supermercado) -> Result<Vec<String>, anyhow::Er
         Supermercado::Jumbo => sites::jumbo::get_urls().await?,
         Supermercado::Carrefour => sites::carrefour::get_urls().await?,
         Supermercado::Coto => sites::coto::get_urls().await?,
+        Supermercado::Farmacity => sites::farmacity::get_urls().await?,
     })
 }
 
@@ -305,6 +306,9 @@ async fn scrap_url(
             sites::coto::parse(url, &tl::parse(body, tl::ParserOptions::default())?)
         }
         "www.jumbo.com.ar" => sites::jumbo::scrap(client, url, body).await,
+        "www.farmacity.com" => {
+            sites::farmacity::parse(url, &tl::parse(body, tl::ParserOptions::default())?)
+        }
         s => bail!("Unknown host {}", s),
     }
 }
diff --git a/scraper-rs/src/sites/farmacity.rs b/scraper-rs/src/sites/farmacity.rs
new file mode 100644
index 0000000..5e9cc11
--- /dev/null
+++ b/scraper-rs/src/sites/farmacity.rs
@@ -0,0 +1,50 @@
+use anyhow::Context;
+use simple_error::bail;
+
+use crate::sites::common;
+use crate::PrecioPoint;
+
+use super::vtex;
+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,
+    })
+}
+
+pub async fn get_urls() -> anyhow::Result<Vec<String>> {
+    let urls = vec![
+        "https://www.farmacity.com/sitemap/product-0.xml",
+        "https://www.farmacity.com/sitemap/product-1.xml",
+    ];
+    vtex::get_urls_from_sitemap(urls).await
+}
diff --git a/scraper-rs/src/sites/mod.rs b/scraper-rs/src/sites/mod.rs
index e305f94..d96e38e 100644
--- a/scraper-rs/src/sites/mod.rs
+++ b/scraper-rs/src/sites/mod.rs
@@ -2,5 +2,6 @@ pub mod carrefour;
 mod common;
 pub mod coto;
 pub mod dia;
+pub mod farmacity;
 pub mod jumbo;
 pub mod vtex;
diff --git a/scraper-rs/src/supermercado.rs b/scraper-rs/src/supermercado.rs
index d7cdbc0..19bbecb 100644
--- a/scraper-rs/src/supermercado.rs
+++ b/scraper-rs/src/supermercado.rs
@@ -6,6 +6,7 @@ pub enum Supermercado {
     Jumbo,
     Carrefour,
     Coto,
+    Farmacity,
 }
 impl Supermercado {
     pub fn host(&self) -> &'static str {
@@ -14,6 +15,7 @@ impl Supermercado {
             Self::Carrefour => "www.carrefour.com.ar",
             Self::Coto => "www.cotodigital3.com.ar",
             Self::Jumbo => "www.jumbo.com.ar",
+            Self::Farmacity => "www.farmacity.com",
         }
     }
 }
diff --git a/sitio/src/routes/ean/[ean]/+page.svelte b/sitio/src/routes/ean/[ean]/+page.svelte
index d3ce771..78cc6a8 100644
--- a/sitio/src/routes/ean/[ean]/+page.svelte
+++ b/sitio/src/routes/ean/[ean]/+page.svelte
@@ -18,6 +18,7 @@
     [Supermercado.Carrefour]: "bg-[#19549d] focus:ring-[#19549d]",
     [Supermercado.Coto]: "bg-[#e20025] focus:ring-[#e20025]",
     [Supermercado.Jumbo]: "bg-[#2dc850] focus:ring-[#2dc850]",
+    [Supermercado.Farmacity]: "bg-[#EF7603] focus:ring-[#EF7603]",
   };
 </script>