sarafg11

Untitled

Dec 18th, 2019
68
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 4.41 KB | None | 0 0
  1. {
  2. "start_urls":
  3. [
  4. "https://www.amazon.es/b/?node=6347717031&ref_=Oct_CateC_6198072031_6347717031&pf_rd_p=7176fd55-a35c-58f4-bf6f-f65010ce11ed&pf_rd_s=merchandised-search-4&pf_rd_t=101&pf_rd_i=6198072031&pf_rd_m=A1AT7YVPFBWXBL&pf_rd_r=9XQ0HVDRZJDS1EWDKZPW&pf_rd_r=9XQ0HVDRZJDS1EWDKZPW&pf_rd_p=7176fd55-a35c-58f4-bf6f-f65010ce11ed",
  5. "https://www.amazon.com/s?i=specialty-aps&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A11058281&ref=nav_em_T1_0_4_NaN_1__nav_desktop_sa_intl_makeup",
  6. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A11060451&dc&fst=as%3Aoff&qid=1573661402&rnid=16225006011&ref=lp_16225006011_nr_n_1",
  7. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A11057241&dc&fst=as%3Aoff&qid=1573661437&rnid=16225006011&ref=lp_16225006011_nr_n_2",
  8. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A11056591&dc&fst=as%3Aoff&qid=1573661463&rnid=16225006011&ref=lp_16225006011_nr_n_3",
  9. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A17242866011&dc&fst=as%3Aoff&qid=1573661476&rnid=16225006011&ref=lp_16225006011_nr_n_4",
  10. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A11062741&dc&fst=as%3Aoff&qid=1573661490&rnid=16225006011&ref=lp_16225006011_nr_n_5",
  11. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A3778591&dc&fst=as%3Aoff&qid=1573661502&rnid=16225006011&ref=lp_16225006011_nr_n_6",
  12. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A3777891&dc&fst=as%3Aoff&qid=1573661569&rnid=16225006011&ref=lp_16225006011_nr_n_7",
  13. "https://www.amazon.com/s?i=beauty-intl-ship&bbn=16225006011&rh=n%3A%2116225006011%2Cn%3A10079992011&dc&fst=as%3Aoff&qid=1573661580&rnid=16225006011&ref=lp_16225006011_nr_n_8",
  14. "https://www.amazon.es/s?bbn=2928521031&rh=n%3A2454136031%2Cn%3A%212665403031%2Cn%3A2928521031%2Cp_89%3ALe+Coq+Sportif&dc&fst=as%3Aoff&qid=1574846475&rnid=1692911031&ref=sr_in_l_p_89_113",
  15. "https://www.amazon.es/s?bbn=2928484031&rh=n%3A2454136031%2Cn%3A%212665403031%2Cn%3A2928484031%2Cp_89%3ALe+Coq+Sportif&dc&fst=as%3Aoff&qid=1574846560&rnid=1692911031&ref=sr_in_l_p_89_37",
  16. "https://www.amazon.es/s?k=KIN&ref=bl_dp_s_web_0",
  17. "https://www.amazon.es/s?i=hpc&bbn=3677430031&rh=n%3A3677430031%2Cp_89%3AKIN&dc&page=4&fst=as%3Aoff&qid=1574872912&rnid=1692911031&ref=sr_pg_1",
  18. "https://www.amazon.es/s?bbn=3677430031&rh=n%3A3677430031%2Cp_89%3AKin+s.a.&dc&fst=as%3Aoff&qid=1574872838&rnid=1692911031&ref=sr_in_k_p_89_542",
  19. "https://www.amazon.es/s?k=KIN+SA.+LABORATORIO+-+BENJAMIN&ref=bl_dp_s_web_0"
  20. ],
  21. "fields": {
  22. "name": "//h1[@id='title']/span[@id='productTitle']/text()",
  23. "price": "//span[@id='priceblock_ourprice']/text()",
  24. "price_old": "//span[@class='priceBlockStrikePriceString a-text-strike']/text()",
  25. "reference": "",
  26. "image": "//div[@id='imgTagWrapperId']/img/@data-a-dynamic-image",
  27. "description": "//div[@id='productDescription']//text()",
  28. "brand": "",
  29. "brand_text": "//*[@id='bylineInfo_feature_div']/div/a/text()",
  30. "categories": "//div[@id='wayfinding-breadcrumbs_container']//ul/li/span[@class='a-list-item']/a/text()",
  31. "availability": "//div[@class='a-section']/div[@id='availability']/span/text()",
  32. "rating_average": "//span[@class='a-declarative']//span[@data-hook='rating-out-of-text']/text()",
  33. "rating_count": "//*[@id='reviewsMedley']/div/div[1]/div[1]/div[2]/span/text()",
  34. "barcode": "",
  35. "stock": "",
  36. "seller": "//div[@id='merchant-info']/a[@id='sellerProfileTriggerId']/text()"
  37. },
  38. "links": {
  39. "detail": "//h2/a[contains(@class,'a-text-normal') and not(contains(@href,'pantry'))]|//div[@class='s-item-container' and not(self::*//img[contains(@src,'pantry')])]//h2/parent::a[contains(@class,'a-text-normal')]",
  40. "next_page": "//a[(contains(.,'Siguiente') or contains(.,'siguiente')) and not(contains(@class,'carousel-goto-nextpage'))]"
  41. },
  42. "regex": {
  43. "rating_average": ["(.*?)de", 1],
  44. "rating_count": ["(.*?)valoraci", 1]
  45. },
  46. "list": {
  47. "product": "//div[@id='mainResults']//li[contains(@id, 'result')]/div | //span[@data-component-type='s-search-results']//div[@class='sg-col-inner']",
  48. "page": " //div[@id='pagn']/span[@class='pagnCur']/text()| //ul[@class='a-pagination']/li[@class='a-selected']//text()"
  49. }
  50. }
Add Comment
Please, Sign In to add comment