sogou_videos.py 2.2 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980
  1. # SPDX-License-Identifier: AGPL-3.0-or-later
  2. """Sogou-Videos: A search engine for retrieving videos from Sogou."""
  3. from urllib.parse import urlencode
  4. from datetime import datetime, timedelta
  5. from searx.exceptions import SearxEngineAPIException
  6. about = {
  7. "website": "https://v.sogou.com/",
  8. "use_official_api": False,
  9. "require_api_key": False,
  10. "results": "JSON",
  11. }
  12. categories = ["videos"]
  13. paging = True
  14. results_per_page = 10
  15. # Base URL
  16. base_url = "https://v.sogou.com"
  17. def request(query, params):
  18. query_params = {
  19. "page": params["pageno"],
  20. "pagesize": 10,
  21. "query": query,
  22. }
  23. params["url"] = f"{base_url}/api/video/shortVideoV2?{urlencode(query_params)}"
  24. return params
  25. def response(resp):
  26. try:
  27. data = resp.json()
  28. except Exception as e:
  29. raise SearxEngineAPIException(f"Invalid response: {e}") from e
  30. results = []
  31. if not data.get("data", {}).get("list"):
  32. raise SearxEngineAPIException("Invalid response")
  33. for entry in data["data"]["list"]:
  34. if not entry.get("titleEsc") or not entry.get("url"):
  35. continue
  36. video_url = entry.get("url")
  37. if video_url.startswith("/vc/np"):
  38. video_url = f"{base_url}{video_url}"
  39. published_date = None
  40. if entry.get("date") and entry.get("duration"):
  41. try:
  42. published_date = datetime.strptime(entry['date'], "%Y-%m-%d")
  43. except (ValueError, TypeError):
  44. published_date = None
  45. length = None
  46. if entry.get("date") and entry.get("duration"):
  47. try:
  48. timediff = datetime.strptime(entry['duration'], "%M:%S")
  49. length = timedelta(minutes=timediff.minute, seconds=timediff.second)
  50. except (ValueError, TypeError):
  51. length = None
  52. results.append(
  53. {
  54. 'url': video_url,
  55. 'title': entry["titleEsc"],
  56. 'content': entry['site'],
  57. 'length': length,
  58. 'template': 'videos.html',
  59. 'publishedDate': published_date,
  60. 'thumbnail': entry["picurl"],
  61. }
  62. )
  63. return results