sarafg11

Untitled

Nov 17th, 2020 (edited)
82
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 1.71 KB | None | 0 0
  1. {
  2. "sitemap_urls":
  3. [
  4. "https://www.druni.es/sitemap.xml"
  5. ],
  6. "fields": {
  7. "name": "//span[@class='page-title' and @itemprop='name']/text()",
  8. "price": "(//span[contains(@id,'product-price-')]/span[@class='price']/text())[1]",
  9. "price_old": "(//span[contains(@id,'old-price-')]/span[@class='price']/text())[1]",
  10. "reference": "",
  11. "image": "//meta[@property='og:image']/@content",
  12. "description": "//div[@id='description']//text()",
  13. "brand": "",
  14. "brand_text": "//div[@class='product-brand']/div[contains(@class,'manufacturer')]/a[@class='value']/text()",
  15. "categories": "//div[@class='breadcrumbs']/ul/li[contains(@class,'category')]/a/text()",
  16. "availability": "//script[@type='application/ld+json' and contains(.,'availability')]/text()",
  17. "rating_average": "//script[@type='application/ld+json' and contains(.,'ratingValue')]/text()",
  18. "rating_count": "//script[@type='application/ld+json' and contains(.,'ratingCount')]/text()",
  19. "barcode": "",
  20. "stock": "",
  21. "variation": "//span[@class='product-item-name']/text()"
  22. },
  23. "links": {
  24. "detail": "//div[@class='product-item-info']//a[@class='product-item-link']",
  25. "next_page": "//link[@rel='next']"
  26. },
  27. "variations": {
  28. "product_variation": "//div[@class='item']"
  29. },
  30. "parser_modules": [
  31. "multiple_product_variations_xpath"
  32. ],
  33. "pipelines": ["UrlVariationPipeline"],
  34. "regex": {
  35. "availability": ["availability\\\":\\s\\\"(.*?)\\\"", 1],
  36. "rating_average": ["ratingValue\\\":\\\"(.*?)\\\"", 1],
  37. "rating_count": ["ratingCount\\\":\\\"(.*?)\\\"", 1],
  38. "price": ["(\\d*,\\d{1,2})", 1],
  39. "price_old": ["(\\d*,\\d{1,2})", 1]
  40. },
  41. "sitemap_rules": ["https://(.*)/(.*)"]
  42. }
Advertisement
Add Comment
Please, Sign In to add comment