json.dump(sub_to_shot[show][episode_name], fp1) # plot to shot mapping plot_to_shot = {} for show in list_of_shows: plot_to_shot[show] = {} for show in list_of_shows: for episode_name in video_file_names[show]: # storing the processed part in json plot_to_shot_path = DIR_PLOTSHOT[show]+'/'+episode_name+'_proc_plotshot.json' try: with open(plot_to_shot_path, 'r') as fp1: plot_to_shot[show][episode_name] = json.load(fp1) except IOError: plot_to_shot[show][episode_name] = preprocessor.plot_shot_assigner(plot_to_sub[show][episode_name], sub_to_shot[show][episode_name]) with open(plot_to_shot_path, 'w') as fp1: json.dump(plot_to_shot[show][episode_name], fp1) # fetch transcripts # fetching transcripts depends on the availability of one... action_stamps = {} for show in list_of_shows: action_stamps[show] = {} # get list of directory files for show in list_of_shows: # check whether transcripts exist? dir_entries = [i[:-4] for i in os.listdir(DIR_TRANSC[show]) if i.endswith('.txt')] for f in dir_entries: # got rid of txt suffix # storing the processed part in json
sub_to_shot[index] = json.load(fp1) except IOError: with open(sub_to_shot_path, "w") as fp1: sub_to_shot[index] = sub_shot_assigner(sub_stamps[index], scene_stamps[index]) json.dump(sub_to_shot[index], fp1) plot_to_shot = [None for i in range(no_episodes)] for index, vid_file in enumerate(file_names): # storing the processed part in json plot_to_shot_path = DIR_PLOTSHOT + "/" + vid_file + "_proc_plotshot.json" try: with open(plot_to_shot_path, "r") as fp1: plot_to_shot[index] = json.load(fp1) except IOError: with open(plot_to_shot_path, "w") as fp1: plot_to_shot[index] = plot_shot_assigner(plot_to_sub[index], sub_to_shot[index]) json.dump(plot_to_shot[index], fp1) shot_timestamps, video_descr, shots_list = None, None, None episode_num, query = 1, "jimmy" shot_timestamps, video_descr, shots_list = similarity_fn1( time_stamps[episode_num - 1], sub_to_shot[episode_num - 1], idf[episode_num - 1], tf_idf[episode_num - 1], plot_sentences[episode_num - 1], plot_to_sub[episode_num - 1], sub_text[episode_num - 1], untouched_sub_text[episode_num - 1], plot_to_shot[episode_num - 1], query,