sogou_videos.py 2.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081
  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. "language": "zh",
  12. }
  13. categories = ["videos"]
  14. paging = True
  15. results_per_page = 10
  16. # Base URL
  17. base_url = "https://v.sogou.com"
  18. def request(query, params):
  19. query_params = {
  20. "page": params["pageno"],
  21. "pagesize": 10,
  22. "query": query,
  23. }
  24. params["url"] = f"{base_url}/api/video/shortVideoV2?{urlencode(query_params)}"
  25. return params
  26. def response(resp):
  27. try:
  28. data = resp.json()
  29. except Exception as e:
  30. raise SearxEngineAPIException(f"Invalid response: {e}") from e
  31. results = []
  32. if not data.get("data", {}).get("list"):
  33. raise SearxEngineAPIException("Invalid response")
  34. for entry in data["data"]["list"]:
  35. if not entry.get("titleEsc") or not entry.get("url"):
  36. continue
  37. video_url = entry.get("url")
  38. if video_url.startswith("/vc/np"):
  39. video_url = f"{base_url}{video_url}"
  40. published_date = None
  41. if entry.get("date") and entry.get("duration"):
  42. try:
  43. published_date = datetime.strptime(entry['date'], "%Y-%m-%d")
  44. except (ValueError, TypeError):
  45. published_date = None
  46. length = None
  47. if entry.get("date") and entry.get("duration"):
  48. try:
  49. timediff = datetime.strptime(entry['duration'], "%M:%S")
  50. length = timedelta(minutes=timediff.minute, seconds=timediff.second)
  51. except (ValueError, TypeError):
  52. length = None
  53. results.append(
  54. {
  55. 'url': video_url,
  56. 'title': entry["titleEsc"],
  57. 'content': entry['site'],
  58. 'length': length,
  59. 'template': 'videos.html',
  60. 'publishedDate': published_date,
  61. 'thumbnail': entry["picurl"],
  62. }
  63. )
  64. return results