sarafg11

Untitled

Nov 10th, 2020
78
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 2.17 KB | None | 0 0
  1. {
  2. "start_urls": [
  3. "https://www.europharmaonline.com/bebes-y-recien-nacidos/",
  4. "https://www.europharmaonline.com/infantil/",
  5. "https://www.europharmaonline.com/higiene-y-bano/",
  6. "https://www.europharmaonline.com/problemas-de-salud-adultos/",
  7. "https://www.europharmaonline.com/dietetica-y-nutricion/",
  8. "https://www.europharmaonline.com/cosmetica/",
  9. "https://www.europharmaonline.com/aparatos-electronicos-y-test/",
  10. "https://www.europharmaonline.com/salud-dental/",
  11. "https://www.europharmaonline.com/optica/",
  12. "https://www.europharmaonline.com/ortopedia/",
  13. "https://www.europharmaonline.com/salud-sexual/",
  14. "https://www.europharmaonline.com/fitoterapia/",
  15. "https://www.europharmaonline.com/natural-bio/",
  16. "https://www.europharmaonline.com/vitaminas/",
  17. "https://www.europharmaonline.com/botiquin/",
  18. "https://www.europharmaonline.com/homeopatia/",
  19. "https://www.europharmaonline.com/medicamentos/",
  20. "https://www.europharmaonline.com/promo/"
  21. ],
  22. "fields": {
  23. "name": "//div[contains(@class,'titulo_producto')]/h1/text()",
  24. "price": "substring-before((//table[@id='miyazaki']//td[@data-th='Precio'])[last()]/text(),'/')",
  25. "price_old": "substring-before(//table[@id='miyazaki']//td[@data-th='Antes']/text(),'/')",
  26. "reference": "//td[contains(.,'Ref')]/following-sibling::td//text()",
  27. "image": "//meta[@property='og:image']/@content",
  28. "description": "//div[contains(@class,'descripcion_producto')]//text()",
  29. "brand": "",
  30. "brand_text": "//td[contains(.,'Marca:')]/following-sibling::td//text()",
  31. "categories": "(//div[@id='fil_ariadna']//a)[position()>1]/text()",
  32. "availability": "//script[@type='application/ld+json']/text()",
  33. "rating_average": "(//meta[@itemprop='ratingValue']/@content)[position()>1]//text()",
  34. "rating_count": "//span[@itemprop='reviewCount']/text()",
  35. "barcode": "",
  36. "stock": ""
  37. },
  38. "regex": {
  39. "availability": ["availability\\\":\\\"(.*)\\\", \\\"url", 1]
  40. },
  41. "links": {
  42. "detail": "//div[@class='titulo_producto']/a",
  43. "next_page": "//div[contains(@class, 'navega_paginas')]//li[a[contains(@class, 'fa-arrow-right')]]/a"
  44. }
  45. }
Advertisement
Add Comment
Please, Sign In to add comment