plot_to_sub[show] = {}
idf[show] = {}
tf_idf[show] = {}
for show in list_of_shows:
for episode_name in video_file_names[show]:
plot_to_sub_path = DIR_PLTSUB[show]+'/'+episode_name+'_proc_pltsub.json'
idf_path = DIR_PLTSUB[show]+'/'+episode_name+'_idf.json'
tf_idf_path = DIR_PLTSUB[show]+'/'+episode_name+'_tf_idf.json'
try:
with open(plot_to_sub_path, 'r') as fp1, open(idf_path, 'r') as fp2, open(tf_idf_path, 'r') as fp3:
plot_to_sub[show][episode_name] = json.load(fp1)
idf[show][episode_name] = json.load(fp2)
tf_idf[show][episode_name] = {int(k):v for k,v in json.load(fp3).items()}
except IOError:
result = preprocessor.plot_sub_assigner(plot_sentences[show][episode_name], sub_text[show][episode_name])
plot_to_sub[show][episode_name] = result['plot_to_sub']
idf[show][episode_name] = result['idf']
tf_idf[show][episode_name] = result['tf_idf']
with open(plot_to_sub_path, 'w') as fp1, open(idf_path, 'w') as fp2, open(tf_idf_path, 'w') as fp3:
json.dump(plot_to_sub[show][episode_name], fp1)
json.dump(idf[show][episode_name], fp2)
json.dump(tf_idf[show][episode_name], fp3)
# subtitle to shot mapping
sub_to_shot = {}
for show in list_of_shows:
sub_to_shot[show] = {}
for show in list_of_shows:
for episode_name in video_file_names[show]:
# will work for the first time
plot_to_sub = [None for i in range(no_episodes)]
idf = [None for i in range(no_episodes)]
tf_idf = [None for i in range(no_episodes)]
for index, vid_file in enumerate(file_names):
plot_to_sub_path = DIR_PLTSUB + "/" + vid_file + "_proc_pltsub.json"
idf_path = DIR_PLTSUB + "/" + vid_file + "_idf.json"
tf_idf_path = DIR_PLTSUB + "/" + vid_file + "_tf_idf.json"
try:
with open(plot_to_sub_path, "r") as fp1, open(idf_path, "r") as fp2, open(tf_idf_path, "r") as fp3:
plot_to_sub[index] = json.load(fp1)
idf[index] = json.load(fp2)
tf_idf[index] = {int(k): v for k, v in json.load(fp3).items()}
except IOError:
with open(plot_to_sub_path, "w") as fp1, open(idf_path, "w") as fp2, open(tf_idf_path, "w") as fp3:
t1, t2, t3 = plot_sub_assigner(plot_sentences[index], sub_text[index])
plot_to_sub[index], idf[index], tf_idf[index] = t1, t2, t3
json.dump(plot_to_sub[index], fp1)
json.dump(idf[index], fp2)
json.dump(tf_idf[index], fp3)
sub_to_shot = [None for i in range(no_episodes)]
for index, vid_file in enumerate(file_names):
# storing the processed part in json
sub_to_shot_path = DIR_SUBSHOT + "/" + vid_file + "_proc_subshot.json"
try:
with open(sub_to_shot_path, "r") as fp1:
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])
