Fix some websites scrapers (#99)

* chore: add clean command to package.json file

* feat(scraper): add URL to several scrapers

* fix(scraper): add missing field

package.json

@@ -10,7 +10,8 @@
     "preview": "vite preview",
     "lint": "eslint src -c ./.eslintrc.cjs --ext .jsx,.js,.ts,.tsx",
     "check-types": "tsc --noEmit",
-    "prepare": "husky"
+    "prepare": "husky",
+    "clean": "rm -rf dist"
   },
   "prettier": "@rows/prettier-config",
   "dependencies": {

src/scrappers/babymarket-list-products.yml

@@ -6,6 +6,10 @@ elementParser:
     query: '.product__brand'
     type: text
 
+  - title: Item model
+    query: '.product__model'
+    type: text
+
   - title: Item Description
     query: '.product__title'
     type: text
@@ -17,3 +21,7 @@ elementParser:
   - title: Price
     query: '.product__price'
     type: text
+
+  - title: URL
+    query: .product__link
+    type: link

src/scrappers/babyone.yml

@@ -7,6 +7,10 @@ elementParser:
     query: .product-name
     type: text
 
+  - title: Item model
+    query: 'product-model'
+    type: text
+
   - title: Item brand
     query: .product-brand
     type: text
@@ -17,4 +21,8 @@ elementParser:
 
   - title: Price
     query: .sale-price
-    type: text
+    type: text
+
+  - title: URL
+    query: .product-link
+    type: link

src/scrappers/babypark-de-list-products.yml

@@ -3,6 +3,14 @@ url:
 listElementsQuery: '#kuLandingProductsListUl > li'
 elementParser:
 
+  - title: Item brand
+    query: 'kuBrand'
+    type: text
+
+  - title: Item model
+    query: 'kuModel'
+    type: text
+
   - title: Item Description
     query: '.kuName'
     type: text
@@ -13,4 +21,8 @@ elementParser:
 
   - title: Price
     query: '.kuSalePrice'
-    type: text
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/babypark-nl-list-products.yml

@@ -2,7 +2,14 @@ url:
   - https://www.babypark.nl/search/*
 listElementsQuery: '#kuLandingProductsListUl > li'
 elementParser:
+  - title: Item brand
+    query: 'kuBrand'
+    type: text
 
+  - title: Item model
+    query: 'kuModel'
+    type: text
+
   - title: Item Description
     query: '.kuName'
     type: text
@@ -13,4 +20,8 @@ elementParser:
 
   - title: Price
     query: '.kuSalePrice'
-    type: text
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/bebe9.yml

@@ -2,13 +2,16 @@ url:
   - https://www.bebe9.com/*
 listElementsQuery: '.products-list > article'
 elementParser:
+  - title: Item brand
+    query: .product-brand
+    type: text
 
-  - title: Item Description
-    query: .product-name
+  - title: Item model
+    query: '.product-model'
     type: text
 
-  - title: Item brand
-    query: .product-brand
+  - title: Item Description
+    query: .product-name
     type: text
 
   - title: RRP
@@ -17,4 +20,8 @@ elementParser:
 
   - title: Price
     query: .product-price
-    type: text
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/bol.yml

@@ -2,13 +2,16 @@ url:
   - https://www.bol.com/*
 listElementsQuery: .product-item, .product-item--row
 elementParser:
+  - title: Item brand
+    query: '.product-brand'
+    type: text
 
-  - title: Item Description
-    query: .product-title
+  - title: Item model
+    query: '.product-model'
     type: text
 
-  - title: Brand
-    query: '[itemprop="name"]'
+  - title: Item Description
+    query: .product-title
     type: text
 
   - title: RRP
@@ -17,4 +20,8 @@ elementParser:
 
   - title: Price
     query: .promo-price
-    type: text
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/cdiscount.yml

@@ -1,7 +1,14 @@
 url: 
   - https://www.cdiscount.com/*
-listElementsQuery: '[data-cs-override-id="Offres-LP"] > li'
+listElementsQuery: ul > li[data-sku]
 elementParser:
+  - title: Item brand
+    query: '.brand'
+    type: text
+
+  - title: Item model
+    query: '.model'
+    type: text
 
   - title: Item Description
     query: h2
@@ -12,5 +19,9 @@ elementParser:
     type: text
 
   - title: Price
-    query: .c-price
-    type: text
+    query: .prdtBILPrice > .price
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/dreambaby.yml

@@ -2,11 +2,26 @@ url:
   - https://www.dreambaby.be/*
 listElementsQuery: '[data-name="Product"]'
 elementParser:
+  - title: Item brand
+    query: '.brand'
+    type: text
+
+  - title: Item model
+    query: '.model'
+    type: text
 
   - title: Item Description
     query: 'h3'
     type: text
 
+  - title: RRP
+    query: 'empty'
+    type: text
+
   - title: Price
     query: '[itemprop="price"]'
-    type: text
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/johnlewis.yml

@@ -2,13 +2,16 @@ url:
   - https://www.johnlewis.com/*
 listElementsQuery: article
 elementParser:
+  - title: Item brand
+    query: '[data-test="product-title"] > h2 > span:nth-child(1)'
+    type: text
 
-  - title: Item Description
-    query: '[data-test="product-title"] > h2 > span:nth-child(2)'
+  - title: Item model
+    query: '.model'
     type: text
 
-  - title: Item brand
-    query: '[data-test="product-title"] > h2 > span:nth-child(1)'
+  - title: Item Description
+    query: '[data-test="product-title"] > h2 > span:nth-child(2)'
     type: text
 
   - title: RRP
@@ -17,4 +20,8 @@ elementParser:
 
   - title: Price
     query: '[data-test="product-card-price-now"]'
-    type: text
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/jollyroom.yml

@@ -3,6 +3,14 @@ url:
 listElementsQuery: '.product-info'
 elementParser:
 
+  - title: Item brand
+    query: '[itemprop="brand"]'
+    type: text
+
+  - title: Item model
+    query: '[itemprop="model"]'
+    type: text
+
   - title: Item Description
     query: h3
     type: text
@@ -13,4 +21,8 @@ elementParser:
 
   - title: Price
     query: .price
-    type: text
+    type: text
+
+  - title: URL
+    query: a
+    type: link

src/scrappers/pinkorblue-list-products-it.yml

@@ -6,6 +6,10 @@ elementParser:
     query: '.product__brand'
     type: text
 
+  - title: Item model
+    query: '.product__model'
+    type: text
+
   - title: Item Description
     query: '.product__title'
     type: text
@@ -16,4 +20,8 @@ elementParser:
 
   - title: Price
     query: '.product__price'
-    type: text
+    type: text
+
+  - title: URL
+    query: .product__link
+    type: link

src/scrappers/pinkorblue-list-products-nl.yml

@@ -6,6 +6,10 @@ elementParser:
     query: '.product__brand'
     type: text
 
+  - title: Item model
+    query: '.product__model'
+    type: text
+
   - title: Item Description
     query: '.product__title'
     type: text
@@ -16,4 +20,8 @@ elementParser:
 
   - title: Price
     query: '.product__price'
-    type: text
+    type: text
+
+  - title: URL
+    query: .product__link
+    type: link

src/scrappers/pinkorblue-list-products.yml

@@ -6,6 +6,10 @@ elementParser:
     query: '.product__brand'
     type: text
 
+  - title: Item model
+    query: '.product__model'
+    type: text
+
   - title: Item Description
     query: '.product__title'
     type: text
@@ -16,4 +20,8 @@ elementParser:
 
   - title: Price
     query: '.product__price'
-    type: text
+    type: text
+
+  - title: URL
+    query: .product__link
+    type: link

src/scrappers/rosaoazul-es-list-products.yml

@@ -14,6 +14,10 @@ elementParser:
     query: '.product__brand'
     type: text
 
+  - title: Item model
+    query: '.product__model'
+    type: text
+
   - title: Item Description
     query: '.product__title'
     type: text
@@ -24,4 +28,8 @@ elementParser:
 
   - title: Price
     query: '.product__price'
-    type: text
+    type: text
+
+  - title: URL
+    query: .product__link
+    type: link