sarafg11

Untitled

Nov 5th, 2020
69
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 1.55 KB | None | 0 0
  1. {
  2. "start_urls":
  3. [
  4. "https://www.ametllerorigen.com/begudes/cervesa?PS=24&O=OrderByNameASC",
  5. "https://www.ametllerorigen.com/Begudes/Aigua?map=c,c",
  6. "https://www.ametllerorigen.com/begudes/refrescs?PS=24&O=OrderByNameASC",
  7. "https://www.ametllerorigen.com/Rebost/olives-i-snacks",
  8. "https://www.ametllerorigen.com/Rebost/Fruita-seca-i-llavors"
  9. ],
  10. "fields": {
  11. "name": "(//div[contains(@class,'product-name')]/h1/div[contains(@class,'productName')])[1]/text()",
  12. "price": "//script[contains(.,'bestPriceFormated')]/text()",
  13. "price_old": "",
  14. "reference": "",
  15. "image": "//script[contains(.,'image')]/text()",
  16. "description": "(//div[@class='product-description']//text())[2]",
  17. "brand": "",
  18. "brand_text": "//script[contains(.,'productBrandName')]/text()",
  19. "categories": "//div[@class='bread-crumb']/ul/li/a/text()",
  20. "availability": "//head//script[last()]/text()",
  21. "rating_average": "",
  22. "rating_count": "",
  23. "barcode": "//script[contains(.,'productEans')]/text()",
  24. "stock": ""
  25. },
  26. "links": {
  27. "detail": "//div[contains(@class,'product-title')]/a",
  28. "next_page": ""
  29. },
  30. "regex": {
  31. "brand_text": ["productBrandName\\\":\\\"(.*?)\\\"", 1],
  32. "barcode": ["productEans\\\":\\[\\\"(.*?)\\\"", 1],
  33. "image": ["image\\\":\\\"(.*?)\\\"", 1],
  34. "price": ["bestPriceFormated\\\":\\\"(.*?)\\\"", 1],
  35. "availability": ["\\\"available\\\":(.*),\\\"displayMode", 1]
  36. },
  37. "list": {
  38. "product": "//div[@class='product-container']",
  39. "page": ""
  40. }
  41. }
Add Comment
Please, Sign In to add comment