logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://hacktivis.me/git/searx.git

scanr_structures.py (2039B)


  1. """
  2. ScanR Structures (Science)
  3. @website https://scanr.enseignementsup-recherche.gouv.fr
  4. @provide-api yes (https://scanr.enseignementsup-recherche.gouv.fr/api/swagger-ui.html)
  5. @using-api yes
  6. @results JSON
  7. @stable yes
  8. @parse url, title, content, img_src
  9. """
  10. from json import loads, dumps
  11. from searx.utils import html_to_text
  12. # engine dependent config
  13. categories = ['science']
  14. paging = True
  15. page_size = 20
  16. # search-url
  17. url = 'https://scanr.enseignementsup-recherche.gouv.fr/'
  18. search_url = url + 'api/structures/search'
  19. # do search-request
  20. def request(query, params):
  21. params['url'] = search_url
  22. params['method'] = 'POST'
  23. params['headers']['Content-type'] = "application/json"
  24. params['data'] = dumps({"query": query,
  25. "searchField": "ALL",
  26. "sortDirection": "ASC",
  27. "sortOrder": "RELEVANCY",
  28. "page": params['pageno'],
  29. "pageSize": page_size})
  30. return params
  31. # get response from search-request
  32. def response(resp):
  33. results = []
  34. search_res = loads(resp.text)
  35. # return empty array if there are no results
  36. if search_res.get('total', 0) < 1:
  37. return []
  38. # parse results
  39. for result in search_res['results']:
  40. if 'id' not in result:
  41. continue
  42. # is it thumbnail or img_src??
  43. thumbnail = None
  44. if 'logo' in result:
  45. thumbnail = result['logo']
  46. if thumbnail[0] == '/':
  47. thumbnail = url + thumbnail
  48. content = None
  49. if 'highlights' in result:
  50. content = result['highlights'][0]['value']
  51. # append result
  52. results.append({'url': url + 'structure/' + result['id'],
  53. 'title': result['label'],
  54. # 'thumbnail': thumbnail,
  55. 'img_src': thumbnail,
  56. 'content': html_to_text(content)})
  57. # return results
  58. return results