sarafg11

Untitled

Dec 18th, 2019
124
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 6.09 KB | None | 0 0
  1. {
  2. "start_urls":
  3. [
  4. "https://www.amazon.es/s?bbn=934359031&rh=n%3A599370031%2Cn%3A%21599371031%2Cn%3A664659031%2Cn%3A934359031%2Cp_n_size_browse-bin%3A9598539031&dc&fst=as%3Aoff&qid=1573032534&rnid=948352031&ref=lp_934359031_nr_p_n_size_browse-bin_0",
  5. "https://www.amazon.es/s?bbn=934359031&rh=n%3A599370031%2Cn%3A%21599371031%2Cn%3A664659031%2Cn%3A934359031%2Cp_n_size_browse-bin%3A9598540031&dc&fst=as%3Aoff&qid=1573032974&rnid=948352031&ref=lp_934359031_nr_p_n_size_browse-bin_1",
  6. "https://www.amazon.es/s?bbn=934359031&rh=n%3A599370031%2Cn%3A%21599371031%2Cn%3A664659031%2Cn%3A934359031%2Cp_n_size_browse-bin%3A9598541031&dc&fst=as%3Aoff&qid=1573033008&rnid=948352031&ref=lp_934359031_nr_p_n_size_browse-bin_2",
  7. "https://www.amazon.es/s?bbn=934359031&rh=n%3A599370031%2Cn%3A%21599371031%2Cn%3A664659031%2Cn%3A934359031%2Cp_n_size_browse-bin%3A9598542031&dc&fst=as%3Aoff&qid=1573033042&rnid=948352031&ref=lp_934359031_nr_p_n_size_browse-bin_3",
  8. "https://www.amazon.es/s?bbn=934359031&rh=n%3A599370031%2Cn%3A%21599371031%2Cn%3A664659031%2Cn%3A934359031%2Cp_n_size_browse-bin%3A9598543031&dc&fst=as%3Aoff&qid=1573033077&rnid=948352031&ref=lp_934359031_nr_p_n_size_browse-bin_4",
  9. "https://www.amazon.es/s?bbn=934359031&rh=n%3A599370031%2Cn%3A%21599371031%2Cn%3A664659031%2Cn%3A934359031%2Cp_n_size_browse-bin%3A9598544031&dc&fst=as%3Aoff&qid=1573033109&rnid=948352031&ref=lp_934359031_nr_p_n_size_browse-bin_5",
  10. "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",
  11. "https://www.amazon.es/b/?node=6347570031&ref_=Oct_CateC_6347509031_9&pf_rd_p=efeed45e-aac0-5ef6-8cf6-c09678cca1f4&pf_rd_s=merchandised-search-4&pf_rd_t=101&pf_rd_i=6347509031&pf_rd_m=A1AT7YVPFBWXBL&pf_rd_r=GVAVASBS3GFDD2DJTR3P&pf_rd_r=GVAVASBS3GFDD2DJTR3P&pf_rd_p=efeed45e-aac0-5ef6-8cf6-c09678cca1f4",
  12. "https://www.amazon.es/b/?node=6347510031&ref_=Oct_CateC_6347509031_4&pf_rd_p=efeed45e-aac0-5ef6-8cf6-c09678cca1f4&pf_rd_s=merchandised-search-4&pf_rd_t=101&pf_rd_i=6347509031&pf_rd_m=A1AT7YVPFBWXBL&pf_rd_r=GVAVASBS3GFDD2DJTR3P&pf_rd_r=GVAVASBS3GFDD2DJTR3P&pf_rd_p=efeed45e-aac0-5ef6-8cf6-c09678cca1f4",
  13. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346878031&dc&fst=as%3Aoff&qid=1573662807&rnid=3677431031&ref=sr_nr_n_1",
  14. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4347676031&dc&fst=as%3Aoff&qid=1573662931&rnid=3677431031&ref=sr_nr_n_2",
  15. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346879031&dc&fst=as%3Aoff&qid=1573662983&rnid=3677431031&ref=sr_nr_n_3",
  16. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346880031&dc&fst=as%3Aoff&qid=1573662957&rnid=3677431031&ref=sr_nr_n_4",
  17. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346882031&dc&fst=as%3Aoff&qid=1573663005&rnid=3677431031&ref=sr_nr_n_5",
  18. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346884031&dc&fst=as%3Aoff&qid=1573663030&rnid=3677431031&ref=sr_nr_n_6",
  19. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4347698031&dc&fst=as%3Aoff&qid=1573663051&rnid=3677431031&ref=sr_nr_n_7",
  20. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346918031&dc&fst=as%3Aoff&qid=1573663095&rnid=3677431031&ref=sr_nr_n_9",
  21. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A6691131031&dc&fst=as%3Aoff&qid=1573663203&rnid=3677431031&ref=sr_nr_n_10",
  22. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346885031&dc&fst=as%3Aoff&qid=1573663602&rnid=3677431031&ref=sr_nr_n_11",
  23. "https://www.amazon.es/s?i=pantry&srs=10547412031&bbn=10858545031&rh=n%3A10858545031%2Cn%3A%213677431031%2Cn%3A4346886031&dc&fst=as%3Aoff&qid=1573663621&rnid=3677431031&ref=sr_nr_n_12"
  24. ],
  25. "fields": {
  26. "name": "//h1[@id='title']/span[@id='productTitle']/text()",
  27. "price": "//td[@class='a-span12']/span[@id='priceblock_ourprice']",
  28. "price_old": "//span[@class='priceBlockStrikePriceString a-text-strike']/text()",
  29. "reference": "",
  30. "image": "//div[@id='imgTagWrapperId']/img/@src",
  31. "description": "//div[@id='productDescription']//text()",
  32. "brand": "",
  33. "brand_text": "//div[@id='merchant-info']/a[@id='sellerProfileTriggerId']/text()",
  34. "categories": "//div[@id='wayfinding-breadcrumbs_container']//ul/li/span[@class='a-list-item']/a/text()",
  35. "availability": "//div[@class='a-section']/div[@id='availability']/span/text()",
  36. "rating_average": "//span[@class='a-declarative']//span[@data-hook='rating-out-of-text']/text()",
  37. "rating_count": "//*[@id='reviewsMedley']/div/div[1]/div[1]/div[2]",
  38. "barcode": "",
  39. "stock": "",
  40. "seller": "//div[@id='merchant-info']/a[@id='sellerProfileTriggerId']/text()"
  41. },
  42. "links": {
  43. "detail": "(//i[@aria-label='Amazon Pantry']/parent::span/parent::a[contains(@class,'a-link-normal')]|//img[@aria-label='Amazon Pantry']/preceding-sibling::a[contains(@class,'a-text-normal')])",
  44. "next_page": "//a[(contains(.,'Siguiente') or contains(.,'siguiente')) and not(contains(@class,'carousel-goto-nextpage'))]"
  45. },
  46. "regex": {
  47. "rating_average": ["(.*?)de", 1],
  48. "rating_count": ["(.*?)valoraciones", 1]
  49. },
  50. "list": {
  51. "product": "//div[@id='mainResults']//li[contains(@id, 'result')]/div | //span[@data-component-type='s-search-results']//div[@class='sg-col-inner']",
  52. "page": " //div[@id='pagn']/span[@class='pagnCur']/text()| //ul[@class='a-pagination']/li[@class='a-selected']//text()"
  53. }
  54. }
Advertisement
Add Comment
Please, Sign In to add comment