sarafg11

Untitled

Oct 14th, 2020 (edited)
93
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 2.09 KB | None | 0 0
  1. {
  2. "start_urls": [
  3. "https://www.notino.es/summer-sale/",
  4. "https://www.notino.es/perfumes-ofertas/",
  5. "https://www.notino.es/perfumes/",
  6. "https://www.notino.es/niche/",
  7. "https://www.notino.es/cosmetica/cosmetica-capilar/",
  8. "https://www.notino.es/dermocosmeticos/",
  9. "https://www.notino.es/cosmetica/cuidado-facial/",
  10. "https://www.notino.es/cosmetica/cuidado-corporal/",
  11. "https://www.notino.es/cosmetica-decorativa/",
  12. "https://www.notino.es/cosmetica-masculina/",
  13. "https://www.notino.es/cosmetica/cuidado-solar/",
  14. "https://www.notino.es/marcas-cosmetica/"
  15. ],
  16. "fields": {
  17. "name": "//div[@id='pdHeader']/h1/span[contains(@class,'styled__Span')]/text()",
  18. "price": "//div[contains(@class,'PriceLabelWrapper')]/span[not(contains(@class,'Currency'))]/text() | //div[@id='pd-price']/span[not(contains(@class,'Currency'))]/text()",
  19. "price_old": "",
  20. "reference": "",
  21. "image": "//meta[@property='og:image']/@content",
  22. "description": "//div[@id='tabAnchor']",
  23. "brand": "",
  24. "brand_text": "//div[@id='pdHeader']/h1/span[contains(@class,'styled__Brand')]/text()",
  25. "categories": "(//div[contains(@class,'styled__BreadcrumbInsideWrapper')]/a)[position()>1]",
  26. "availability": "(//link[@itemprop='availability']/@href)[1]",
  27. "rating_average": "//div[@itemtype='http://schema.org/AggregateRating']//meta[@itemprop='ratingValue']/@content",
  28. "rating_count": "//div[@itemtype='http://schema.org/AggregateRating']//span[@itemprop='ratingCount']/@content",
  29. "barcode": "//img[contains(@class,'MainImg')]/@srcset",
  30. "stock": "",
  31. "variation": "//div[contains(@class,'pd-variant-label')]/text()"
  32. },
  33. "variations": {
  34. "product_variation": "//li[contains(@class,'VariantListItem')]"
  35. },
  36. "parser_modules": [
  37. "multiple_product_variations_xpath"
  38. ],
  39. "links": {
  40. "detail": "//li[@class='item']/a[contains(@class,'spc')]",
  41. "next_page": "//p[@class='paging']/span[@class='pages']/a[@class='next']"
  42. },
  43. "regex": {
  44. "barcode": ["(\\d{12,13})", 1]
  45. },
  46. "pipelines": ["UrlVariationPipeline"]
  47. }
Advertisement
Add Comment
Please, Sign In to add comment