Advertisement
sarafg11

Untitled

Dec 15th, 2020 (edited)
55
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 1.88 KB | None | 0 0
  1. {
  2. "start_urls": [
  3. "https://www.demarpisan.es/c391226-perros.html",
  4. "https://www.demarpisan.es/c199106-gatos.html",
  5. "https://www.demarpisan.es/c391454-parafarmacia-perros-gatos.html",
  6. "https://www.demarpisan.es/c390487-fungibles-varios-veterinaria.html",
  7. "https://www.demarpisan.es/c217814-caballos.html",
  8. "https://www.demarpisan.es/c390701-aves-y-peq-animal.html",
  9. "https://www.demarpisan.es/c261964-promociones-de-verano.html"
  10. ],
  11. "fields": {
  12. "name": "//h1[@id='product-name']/text()",
  13. "price": "(//meta[@property='product:price:amount']/@content|//option[@value]/text())[1]",
  14. "price_old": "",
  15. "reference": "//script[@type='application/ld+json' and contains(.,'sku')]/text()",
  16. "image": "//meta[@property='og:image']/@content",
  17. "description": "//div[contains(@class,'info')]//*//text()",
  18. "brand": "",
  19. "brand_text": "//script[@type='text/javascript' and contains(.,'brand')]/text()",
  20. "categories": "(//div[@id='breadcrumbs']/div/span/a/span)[position()>1]/text()",
  21. "availability": "//script[@type='application/ld+json' and contains(., 'availability')]/text()",
  22. "rating_average": "",
  23. "rating_count": "",
  24. "barcode": "",
  25. "stock": "",
  26. "variation": "substring-before(//option[@value]/text(), ' - ')"
  27. },
  28. "links": {
  29. "detail": "//div[contains(@class,'featured-title')]/a",
  30. "next_page": "//div[@class='art-list-view-pager-nav']//a[@class='next']"
  31. },
  32. "variations": {
  33. "product_variation": "//select[@class='product-options-dropdown']/option"
  34. },
  35. "regex": {
  36. "availability": ["availability\\\":\\\"(.*?)\\\"", 1],
  37. "brand_text": ["brand\\\":\\\"(.*?)\\\"", 1],
  38. "reference": ["sku\\\":\\\"(.*?)\\\"", 1],
  39. "price": ["((\\d+ x )?\\d+ [^\\s]+ - )?(.*)", 3]
  40. },
  41. "parser_modules": [
  42. "multiple_product_variations_xpath"
  43. ],
  44. "pipelines": ["UrlVariationPipeline"]
  45. }
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement