From da9f2c83488dbd8213b2e8ef414af6615cbe2ce5 Mon Sep 17 00:00:00 2001
From: Nulo <git@nulo.in>
Date: Thu, 4 Jan 2024 17:48:27 -0300
Subject: [PATCH] cli: poder scrappear links especificos

---
 scraper/cli.ts   |  7 ++++++-
 scraper/scrap.ts | 18 ++++++++++--------
 2 files changed, 16 insertions(+), 9 deletions(-)

diff --git a/scraper/cli.ts b/scraper/cli.ts
index 3e8af80..b68bda7 100644
--- a/scraper/cli.ts
+++ b/scraper/cli.ts
@@ -2,7 +2,7 @@ import { scrapCarrefourProducts } from "../carrefour-link-scraper/index.js";
 import { scrapCotoProducts } from "../coto-link-scraper/index.js";
 import { scrapDiaProducts } from "../dia-link-scraper/index.js";
 import { auto } from "./auto.js";
-import { downloadList } from "./scrap.js";
+import { downloadList, getProduct } from "./scrap.js";
 
 if (process.argv[2] === "auto") {
   await auto();
@@ -12,6 +12,11 @@ if (process.argv[2] === "auto") {
   await scrapDiaProducts();
 } else if (process.argv[2] === "scrap-coto-links") {
   await scrapCotoProducts();
+} else if (process.argv[2] === "scrap-link") {
+  const url = new URL(process.argv[3]);
+  const res = await fetch(url);
+  const text = await res.text();
+  console.info(getProduct(url, text));
 } else if (process.argv[2] === "scrap") {
   const urlLists = process.argv.slice(3);
   if (urlLists.length > 0) {
diff --git a/scraper/scrap.ts b/scraper/scrap.ts
index 8035a13..f482492 100644
--- a/scraper/scrap.ts
+++ b/scraper/scrap.ts
@@ -60,6 +60,15 @@ export async function downloadList(path: string) {
   return progress;
 }
 
+export function getProduct(url: URL, html: string) {
+  if (url.hostname === "www.carrefour.com.ar") return getCarrefourProduct(html);
+  else if (url.hostname === "diaonline.supermercadosdia.com.ar")
+    return getDiaProduct(html);
+  else if (url.hostname === "www.cotodigital3.com.ar")
+    return getCotoProduct(html);
+  else throw new Error(`Unknown host ${url.hostname}`);
+}
+
 type ScrapResult =
   | { type: "skipped" }
   | { type: "done" }
@@ -81,14 +90,7 @@ async function scrap(urlS: string): Promise<ScrapResult> {
   const html = await res.text();
 
   try {
-    let ish: Precioish | undefined = undefined;
-    if (url.hostname === "www.carrefour.com.ar")
-      ish = getCarrefourProduct(html);
-    else if (url.hostname === "diaonline.supermercadosdia.com.ar")
-      ish = getDiaProduct(html);
-    else if (url.hostname === "www.cotodigital3.com.ar")
-      ish = getCotoProduct(html);
-    else throw new Error(`Unknown host ${url.hostname}`);
+    let ish = getProduct(url, html);
 
     const p: Precio = {
       ...ish,