This commit is contained in:
Cat /dev/Nulo 2024-01-04 19:25:17 -03:00
parent e890d5f63b
commit 525510a8dd
9 changed files with 212 additions and 3 deletions

3
data/Jumbo.txt Normal file
View file

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:6299b470d9debc9a173b40c2ba91208eb43a6c8cde02a4819e7bcd76368e4363
size 922185

100
data/samples/Jumbo.100.txt Normal file
View file

@ -0,0 +1,100 @@
https://www.jumbo.com.ar/huevos-de-color-avicoper-6-u-1-paquete-2/p
https://www.jumbo.com.ar/ajo-ahumado-organico-pampa-gourmet-285g/p
https://www.jumbo.com.ar/boxer-dst-raya-finita-art-b278-talle-m/p
https://www.jumbo.com.ar/yogur-bebible-ser-sachet-vainilla-900g/p
https://www.jumbo.com.ar/plato-playo-melamina-27-cm-boho-krea-2/p
https://www.jumbo.com.ar/mermelada-la-vieja-fabrica-frutos-del-bosque-350-gr/p
https://www.jumbo.com.ar/dr-lemon-vodka-pomelo-5/p
https://www.jumbo.com.ar/vino-cuvelier-los-andes-grand-vin-750cc/p
https://www.jumbo.com.ar/capsulas-cafe-cabrales-dg-cortado-x88gr/p
https://www.jumbo.com.ar/pizza-muzarella-e/p
https://www.jumbo.com.ar/filet-de-merluza-rebozado-8/p
https://www.jumbo.com.ar/ron-bacardi-carta-blanca-750-ml/p
https://www.jumbo.com.ar/sal-gruesa-celusal-1-kg/p
https://www.jumbo.com.ar/vaso-bajo-acrilico-boho-krea-2/p
https://www.jumbo.com.ar/espumante-chandon-demi-sec/p
https://www.jumbo.com.ar/jarra-electrica-smartlife-sl-ek1714wpn/p
https://www.jumbo.com.ar/espumante-dada-7-rose-dulce-750-cc/p
https://www.jumbo.com.ar/panquequera-hudson-de-aluminio-con-antiadherente-22cm/p
https://www.jumbo.com.ar/sacapuntas-de-plastico-pizzini-2un/p
https://www.jumbo.com.ar/vino-vinas-de-alvear-tinto-750ml/p
https://www.jumbo.com.ar/campera-mujer-puffer-larga/p
https://www.jumbo.com.ar/tabla-de-quesos/p
https://www.jumbo.com.ar/frutos-del-bosque-frutas-del-sur-x400gr/p
https://www.jumbo.com.ar/blister-resaltador-flash-amarillo-x-1-un/p
https://www.jumbo.com.ar/alim-whiskas-gatitos-carne-y-leche-500gr/p
https://www.jumbo.com.ar/detergente-polvo-zorro-blue-3k-x-1un/p
https://www.jumbo.com.ar/media-vestir-hombre-1s10471-negro/p
https://www.jumbo.com.ar/nachos-macritas-ketchup-x90g/p
https://www.jumbo.com.ar/pack-x3-medias-juvenil-liso-t-5-elemento/p
https://www.jumbo.com.ar/set-de-vehiculos-emergencias-duravit/p
https://www.jumbo.com.ar/carbon-patagonia-x-4kgs/p
https://www.jumbo.com.ar/rejilla-mr-trapo-cocina-algodon/p
https://www.jumbo.com.ar/jugo-exprimido-pura-frutta-arandanos-manzana-verde-x-1l/p
https://www.jumbo.com.ar/media-dama-invisible-alta-nyb-urb-2/p
https://www.jumbo.com.ar/boxer-nino-raya-violeta-2-colores-dst-t-10/p
https://www.jumbo.com.ar/barra-zafran-caju-y-sem-de-zapallo-x112g/p
https://www.jumbo.com.ar/iniciador-de-fuego-maderasa/p
https://www.jumbo.com.ar/queso-mozzarella-barraza-x-500grs-paq-gr-500/p
https://www.jumbo.com.ar/vaso-de-vidrio-cuadrado-360-cc/p
https://www.jumbo.com.ar/shampoo-sedal-jengibre-y-ricino-190ml/p
https://www.jumbo.com.ar/roller-gel-filgo-gel-pop-glitter-1un/p
https://www.jumbo.com.ar/una-familia-anormal-el-misterio-de-prh/p
https://www.jumbo.com.ar/veggie-stick-tomate-y-oliva-via-vita-x-50grs/p
https://www.jumbo.com.ar/bowl-stor-bicolor-mickey-mouse/p
https://www.jumbo.com.ar/vino-blanco-don-valentin-lacrado-750-ml/p
https://www.jumbo.com.ar/un-vecino-anormal-2-prh/p
https://www.jumbo.com.ar/paleta-pet-cancat-mordillo-ice/p
https://www.jumbo.com.ar/aceitunas-nucete-premium-descarozadas-180-gr/p
https://www.jumbo.com.ar/caja-plastica-6l-teen-boy-pv23-krea-2/p
https://www.jumbo.com.ar/vino-santa-julia-chardonnay-x-750-cc/p
https://www.jumbo.com.ar/protecor-solar-dermaglos-bebes-fps65-120gr/p
https://www.jumbo.com.ar/oregano-100-gr/p
https://www.jumbo.com.ar/puerro-song/p
https://www.jumbo.com.ar/repuesto-difusor-sandia-pepino-350-ml-2/p
https://www.jumbo.com.ar/botellas-plasticas-origin-580ml-rosa-2/p
https://www.jumbo.com.ar/nescafe-dolca-original-x-170gr/p
https://www.jumbo.com.ar/tapa-empanada-veggie-signo-de-oro-x-500g/p
https://www.jumbo.com.ar/inflador-de-pie-bestway-air-hammer/p
https://www.jumbo.com.ar/ketchup-ahumado-marian-arytza-400g/p
https://www.jumbo.com.ar/sal-marina-finas-hierbas-ahumada-s-tacc-450g/p
https://www.jumbo.com.ar/jugo-smudis-pomelo-500ml-brk-0-5-lt/p
https://www.jumbo.com.ar/limpiador-antihongos-ayudin-removedor-activo-envase-economico-450-ml/p
https://www.jumbo.com.ar/marcador-permanente-punta-redonda-color-negro/p
https://www.jumbo.com.ar/galletitas-dulces-con-chips-de-chocolate-pepitos-119g/p
https://www.jumbo.com.ar/afeitadora-bic-comfort-twin-l5p4-2/p
https://www.jumbo.com.ar/canvas-20x20-cm-paisajes-04-krea/p
https://www.jumbo.com.ar/turron-georgalos-de-mani-con-chocolate-x-90-gr/p
https://www.jumbo.com.ar/arroz-vanguardia-elaborado-largo-fino/p
https://www.jumbo.com.ar/set-x-3-pastafrola-fija-n-14/p
https://www.jumbo.com.ar/pulpa-fina-basilico-mutti-400-gr/p
https://www.jumbo.com.ar/vino-tinto-elementos-malbec-750-cc/p
https://www.jumbo.com.ar/enjuague-bucal-listerine-antisarro-suave-sn-alcohol-x250/p
https://www.jumbo.com.ar/almohaditas-lasfor-avellana-200-grs/p
https://www.jumbo.com.ar/vino-tinto-los-haroldos-estate-cabernet-sauvignon-750-ml/p
https://www.jumbo.com.ar/peluche-funnyland-maxtoys-tibalt-perro-28cm/p
https://www.jumbo.com.ar/cafetera-filtro-negro-electrolux-1-2-litros/p
https://www.jumbo.com.ar/media-nina-ciudadella-minnie-t2/p
https://www.jumbo.com.ar/portaretrato-colores-13x18cm-4c-krea4136010100/p
https://www.jumbo.com.ar/lustramuebles-blem-madera-aceite-de-argan-aerosol-360cc/p
https://www.jumbo.com.ar/sriracha-sauce-hashi-x250ml-2/p
https://www.jumbo.com.ar/plato-hondo-22-1-cm-ceramica-blanca/p
https://www.jumbo.com.ar/limpiador-harpic-banos-sarro-y-manchas-495ml/p
https://www.jumbo.com.ar/shampoo-dove-real-poder-de-las-plantas-purificacion-jengibre-300-ml/p
https://www.jumbo.com.ar/aromatizador-glade-mini-gel-car-3/p
https://www.jumbo.com.ar/carpeta-con-10-folios-a4/p
https://www.jumbo.com.ar/sabana-king-caracol-krea/p
https://www.jumbo.com.ar/leche-en-polvo-nutribaby-1-hmo-x-800-grs/p
https://www.jumbo.com.ar/chalitas-viavita-clasicas-x-100-grs-sin-tacc/p
https://www.jumbo.com.ar/hervidor-tramontina-14cm-cm-x1/p
https://www.jumbo.com.ar/aceitunas-de-gordal-ybarra-x240gr-2/p
https://www.jumbo.com.ar/tableta-vizzio-relleno-nugaton-x100g-2/p
https://www.jumbo.com.ar/mortadela-paladini-fetas-finas-x-200-gr-2/p
https://www.jumbo.com.ar/budin-limon-y-amapolas/p
https://www.jumbo.com.ar/vino-chac-chac-sauvingnon-blanc-lata-269cc/p
https://www.jumbo.com.ar/whisky-chivas-regal-18-yo-700cc/p
https://www.jumbo.com.ar/copa-de-vidrio-rigolleau-6/p
https://www.jumbo.com.ar/notcreamcheese-210-gr/p
https://www.jumbo.com.ar/oso-con-miel-de-abejas-cuisine-co-340-gr/p
https://www.jumbo.com.ar/difusor-aromas-spirit-spirit-win-home-250ml-x1/p
https://www.jumbo.com.ar/exprimidor-ultracomb-ex-2302/p

View file

@ -2,15 +2,18 @@ export enum Supermercado {
Dia = "Dia",
Carrefour = "Carrefour",
Coto = "Coto",
Jumbo = "Jumbo",
}
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,
};
export const colorBySupermercado: { [supermercado in Supermercado]: string } = {
[Supermercado.Dia]: "#d52b1e",
[Supermercado.Carrefour]: "#19549d",
[Supermercado.Coto]: "#e20025",
[Supermercado.Jumbo]: "#2dc850",
};

38
link-scrapers/jumbo.ts Normal file
View file

@ -0,0 +1,38 @@
import pMap from "p-map";
import { saveUrls } from "db-datos/urlHelpers.js";
import { getUrlsFromSitemap } from "./common.js";
export async function scrapJumboProducts() {
await scrapBySitemap();
}
async function scrapBySitemap() {
// de https://www.jumbo.com.ar/sitemap.xml
const sitemaps = [
"https://www.jumbo.com.ar/sitemap/product-1.xml",
"https://www.jumbo.com.ar/sitemap/product-10.xml",
"https://www.jumbo.com.ar/sitemap/product-11.xml",
"https://www.jumbo.com.ar/sitemap/product-12.xml",
"https://www.jumbo.com.ar/sitemap/product-13.xml",
"https://www.jumbo.com.ar/sitemap/product-14.xml",
"https://www.jumbo.com.ar/sitemap/product-15.xml",
"https://www.jumbo.com.ar/sitemap/product-2.xml",
"https://www.jumbo.com.ar/sitemap/product-3.xml",
"https://www.jumbo.com.ar/sitemap/product-4.xml",
"https://www.jumbo.com.ar/sitemap/product-5.xml",
"https://www.jumbo.com.ar/sitemap/product-6.xml",
"https://www.jumbo.com.ar/sitemap/product-7.xml",
"https://www.jumbo.com.ar/sitemap/product-8.xml",
"https://www.jumbo.com.ar/sitemap/product-9.xml",
];
await pMap(
sitemaps,
async (sitemapUrl) => {
const res = await fetch(sitemapUrl);
const xml = await res.text();
saveUrls(getUrlsFromSitemap(xml));
},
{ concurrency: 3 }
);
}

View file

@ -11,6 +11,7 @@ import { productoUrls } from "db-datos/schema.js";
import { scrapDiaProducts } from "../link-scrapers/dia.js";
import { scrapCotoProducts } from "../link-scrapers/coto.js";
import { scrapCarrefourProducts } from "../link-scrapers/carrefour.js";
import { scrapJumboProducts } from "../link-scrapers/jumbo.js";
const supermercados: Supermercado[] = [
Supermercado.Carrefour,
@ -59,6 +60,9 @@ class Auto {
case "Carrefour":
await scrapCarrefourProducts();
break;
case "Jumbo":
await scrapJumboProducts();
break;
}
this.inform(
`[scrapUrls[${supermercado}]] Tardó ${formatMs(performance.now() - t0)}`

View file

@ -1,6 +1,7 @@
import { scrapCarrefourProducts } from "../link-scrapers/carrefour.js";
import { scrapCotoProducts } from "../link-scrapers/coto.js";
import { scrapDiaProducts } from "../link-scrapers/dia.js";
import { scrapJumboProducts } from "../link-scrapers/jumbo.js";
import { auto } from "./auto.js";
import { downloadList, getProduct } from "./scrap.js";
@ -12,11 +13,13 @@ if (process.argv[2] === "auto") {
await scrapDiaProducts();
} else if (process.argv[2] === "scrap-coto-links") {
await scrapCotoProducts();
} else if (process.argv[2] === "scrap-jumbo-links") {
await scrapJumboProducts();
} 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));
console.info(await getProduct(url, text));
} else if (process.argv[2] === "scrap") {
const urlLists = process.argv.slice(3);
if (urlLists.length > 0) {

View file

@ -31,6 +31,7 @@ const zProductLd = z.object({
"@type": z.literal("Product"),
name: z.string(),
image: z.string(),
sku: z.string().optional(),
offers: z.object({
offers: z.array(
z.object({

54
scraper/parsers/jumbo.ts Normal file
View file

@ -0,0 +1,54 @@
import { parseHTML } from "linkedom";
import { type Precioish } from "../scrap.js";
import { getProductJsonLd, priceFromMeta, stockFromMeta } from "./common.js";
import { z } from "zod";
const zJumboSearch = z.tuple([
z.object({
items: z.array(
z.object({
ean: z.string(),
})
),
}),
]);
async function getEanFromSearch(sku: string) {
const url = new URL(
"https://www.jumbo.com.ar/api/catalog_system/pub/products/search"
);
url.searchParams.set("fq", `skuId:${sku}`);
const res = await fetch(url);
const json = await res.json();
const parsed = zJumboSearch.parse(json);
const ean = parsed[0].items[0].ean;
if (!parsed[0].items.every((x) => x.ean === ean)) {
throw new Error("Inesperado: no todos los items tienen el mismo EAN");
}
return ean;
}
export async function getJumboProduct(
html: string | Buffer
): Promise<Precioish> {
const dom = parseHTML(html);
const precioCentavos = priceFromMeta(dom);
const inStock = stockFromMeta(dom);
const ld = getProductJsonLd(dom);
const name = ld.name;
const imageUrl = ld.image;
const retailerSku = ld.sku;
if (!retailerSku)
throw new Error("No encontré el SKU de Jumbo para pedir el EAN");
const ean = await getEanFromSearch(retailerSku);
return {
name,
imageUrl,
ean,
precioCentavos,
inStock,
};
}

View file

@ -7,6 +7,7 @@ import { getCotoProduct } from "./parsers/coto.js";
import { join } from "path";
import { db } from "db-datos/db.js";
import pMap from "p-map";
import { getJumboProduct } from "./parsers/jumbo.js";
const DEBUG = true;
const PARSER_VERSION = 4;
@ -60,12 +61,14 @@ export async function downloadList(path: string) {
return progress;
}
export function getProduct(url: URL, html: string) {
export async function getProduct(url: URL, html: string): Promise<Precioish> {
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 if (url.hostname === "www.jumbo.com.ar")
return await getJumboProduct(html);
else throw new Error(`Unknown host ${url.hostname}`);
}
@ -90,7 +93,7 @@ async function scrap(urlS: string): Promise<ScrapResult> {
const html = await res.text();
try {
let ish = getProduct(url, html);
let ish = await getProduct(url, html);
const p: Precio = {
...ish,