filter_video.py 1.1 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546
  1. #coding utf-8
  2. import sys
  3. from utils import filter_video_status
  4. if __name__=="__main__":
  5. f = open(sys.argv[1])
  6. video_set = set('')
  7. video_rec = []
  8. while True:
  9. line = f.readline()
  10. if not line:
  11. break
  12. items = line.strip().split("\t")
  13. if len(items)<3:
  14. continue
  15. vid = -1
  16. try:
  17. vid = int(items[2])
  18. except:
  19. continue
  20. if vid ==-1:
  21. continue
  22. video_set.add(vid)
  23. video_rec.append(line)
  24. f.close()
  25. print(len(video_set))
  26. video_list = list(video_set)
  27. left_video_list = filter_video_status(video_list)
  28. left_video_set = set(left_video_list)
  29. print(left_video_list)
  30. f2 = open(sys.argv[2], 'w')
  31. for line in video_rec:
  32. items = line.strip().split("\t")
  33. if len(items)<3:
  34. continue
  35. vid = -1
  36. try:
  37. vid = int(items[2])
  38. except:
  39. continue
  40. if vid not in left_video_set:
  41. continue
  42. f2.write(line)
  43. f2.close()
  44. #print(len(left_video_list))