sarafg11

Untitled

Aug 14th, 2020
79
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 2.06 KB | None | 0 0
  1. {
  2. "start_urls":
  3. [
  4. "https://www.hipercor.es/supermercado/cuidado-e-higiene-personal/botiquin-y-bienestar-sexual/",
  5. "https://www.hipercor.es/supermercado/cuidado-e-higiene-personal/higiene-personal/",
  6. "https://www.hipercor.es/supermercado/cuidado-e-higiene-personal/maquillaje/",
  7. "https://www.hipercor.es/supermercado/bebes/",
  8. "https://www.hipercor.es/supermercado/parafarmacia/",
  9. "https://www.hipercor.es/supermercado/bebidas/cervezas/?level=1",
  10. "https://www.hipercor.es/supermercado/alimentacion/alimentacion-general/aperitivos-y-frutos-secos/"
  11. ],
  12. "fields": {
  13. "name": "//h3[@class='product_tile-description']/a/text()",
  14. "price": "//div[contains(@class, 'grid-item')]/@data-json",
  15. "price_old": "//div[contains(@class, 'grid-item')]/@data-json",
  16. "image": "//div[@class='product_tile-image _fade ']/a/img/@src",
  17. "brand_text": "//div[contains(@class, 'grid-item')]/@data-json",
  18. "categories": "//div[contains(@class, 'grid-item')]/@data-json",
  19. "availability": "//div[contains(@class, 'grid-item')]/@data-json",
  20. "url": "//h3[@class='product_tile-description']/a/@href"
  21. },
  22. "links": {
  23. "item_detail": {
  24. "detail": "",
  25. "next_page": {
  26. "restrict_xpaths": "//a[@rel='next']",
  27. "callback": "parse_item"
  28. },
  29. "parse_start_url": true
  30. },
  31. "positioning_spider": {
  32. "detail": "//h3[@class='product_tile-description']//a",
  33. "next_page": "//li[@id='pagination-next']/a"
  34. }
  35. },
  36. "regex": {
  37. "price": ["final\\\":\\\"(.*?)\\\"", 1],
  38. "price_old": ["original\\\":\\\"(.*?)\\\"", 1],
  39. "availability": ["status\\\":\\\"(.*?)\\\"", 1],
  40. "brand_text": ["brand\\\":\\\"(.*?)\\\"", 1],
  41. "categories": ["category\\\":.*\\\"(.*?)\\\"\\]", 1]
  42.  
  43. },
  44. "list": {
  45. "product": "//div[contains(@class, 'js-grid')]//div[contains(@class, 'grid-item')]",
  46. "page": "substring-before(substring-after(//li[@id='pagination-current']/text(), 'Página '), ' de')"
  47. },
  48. "parser_modules": [
  49. "multiple_response_xpath",
  50. "multiple_item_xpath"
  51. ]
  52. }
Add Comment
Please, Sign In to add comment