Merge pull request #33 from project-kxkg/SRT_cleanup

Srt cleanup

Former-commit-id: 9a454ba96523d6202b5229ffd620c78a22c3db4b

.gitignore

@@ -1,10 +1,11 @@
-/downloads
-/results
+__pycache__/
 .DS_Store
-/__pycache__
+.idea/
+downloads/
+results/
+test/
 test.py
 test.srt
 test.txt
 log_*.csv
-log.csv
-/test
+log.csv

SRT.py

@@ -1,10 +1,11 @@
-from datetime import timedelta
-from csv import reader
-from datetime import datetime
+import os
 import re
+from copy import copy, deepcopy
+from csv import reader
+from datetime import timedelta
+
 import openai
-import os
-from collections import deque
+
 
 class SRT_segment(object):
     def __init__(self, *args) -> None:
@@ -12,22 +13,24 @@ class SRT_segment(object):
             segment = args[0]
             self.start = segment['start']
             self.end = segment['end']
-            self.start_ms = int((segment['start']*100)%100*10)
-            self.end_ms = int((segment['end']*100)%100*10)
+            self.start_ms = int((segment['start'] * 100) % 100 * 10)
+            self.end_ms = int((segment['end'] * 100) % 100 * 10)
 
-            if self.start_ms == self.end_ms and int(segment['start']) == int(segment['end']): # avoid empty time stamp
-                self.end_ms+=500
+            if self.start_ms == self.end_ms and int(segment['start']) == int(segment['end']):  # avoid empty time stamp
+                self.end_ms += 500
 
             self.start_time = timedelta(seconds=int(segment['start']), milliseconds=self.start_ms)
             self.end_time = timedelta(seconds=int(segment['end']), milliseconds=self.end_ms)
             if self.start_ms == 0:
-                self.start_time_str = str(0)+str(self.start_time).split('.')[0]+',000'
+                self.start_time_str = str(0) + str(self.start_time).split('.')[0] + ',000'
             else:
-                self.start_time_str = str(0)+str(self.start_time).split('.')[0]+','+str(self.start_time).split('.')[1][:3]
+                self.start_time_str = str(0) + str(self.start_time).split('.')[0] + ',' + \
+                                      str(self.start_time).split('.')[1][:3]
             if self.end_ms == 0:
-                self.end_time_str = str(0)+str(self.end_time).split('.')[0]+',000'
+                self.end_time_str = str(0) + str(self.end_time).split('.')[0] + ',000'
             else:
-                self.end_time_str = str(0)+str(self.end_time).split('.')[0]+','+str(self.end_time).split('.')[1][:3]
+                self.end_time_str = str(0) + str(self.end_time).split('.')[0] + ',' + str(self.end_time).split('.')[1][
+                                                                                      :3]
             self.source_text = segment['text'].lstrip()
             self.duration = f"{self.start_time_str} --> {self.end_time_str}"
             self.translation = ""
@@ -39,15 +42,21 @@ class SRT_segment(object):
             self.end_time_str = self.duration.split(" --> ")[1]
 
             # parse the time to float
-            self.start_ms = int(self.start_time_str.split(',')[1])/10
-            self.end_ms = int(self.end_time_str.split(',')[1])/10
+            self.start_ms = int(self.start_time_str.split(',')[1]) / 10
+            self.end_ms = int(self.end_time_str.split(',')[1]) / 10
             start_list = self.start_time_str.split(',')[0].split(':')
-            self.start = int(start_list[0])*3600 + int(start_list[1])*60 + int(start_list[2]) + self.start_ms/100
+            self.start = int(start_list[0]) * 3600 + int(start_list[1]) * 60 + int(start_list[2]) + self.start_ms / 100
             end_list = self.end_time_str.split(',')[0].split(':')
-            self.end = int(end_list[0])*3600 + int(end_list[1])*60 + int(end_list[2]) + self.end_ms/100
+            self.end = int(end_list[0]) * 3600 + int(end_list[1]) * 60 + int(end_list[2]) + self.end_ms / 100
             self.translation = ""
-    
+
     def merge_seg(self, seg):
+        """
+        Merge the segment seg with the current segment in place.
+        :param seg: Another segment that is strictly next to current one.
+        :return: None
+        """
+        # assert seg.start_ms == self.end_ms, f"cannot merge discontinuous segments."
         self.source_text += f' {seg.source_text}'
         self.translation += f' {seg.translation}'
         self.end_time_str = seg.end_time_str
@@ -56,22 +65,42 @@ class SRT_segment(object):
         self.duration = f"{self.start_time_str} --> {self.end_time_str}"
         pass
 
+    def __add__(self, other):
+        """
+        Merge the segment seg with the current segment, and return the new constructed segment.
+        No in-place modification.
+        :param other: Another segment that is strictly next to added segment.
+        :return: new segment of the two sub-segments
+        """
+        # assert other.start_ms == self.end_ms, f"cannot merge discontinuous segments."
+        result = deepcopy(self)
+        result.source_text += f' {other.source_text}'
+        result.translation += f' {other.translation}'
+        result.end_time_str = other.end_time_str
+        result.end = other.end
+        result.end_ms = other.end_ms
+        result.duration = f"{self.start_time_str} --> {self.end_time_str}"
+        return result
+
     def remove_trans_punc(self):
-        # remove punctuations in translation text
-        self.translation = self.translation.replace('，', ' ')
-        self.translation = self.translation.replace('。', ' ')
-        self.translation = self.translation.replace('！', ' ')
-        self.translation = self.translation.replace('？', ' ')
+        """
+        remove punctuations in translation text
+        :return: None
+        """
+        punc_cn = "，。！？"
+        translator = str.maketrans(punc_cn, ' ' * len(punc_cn))
+        self.translation = self.translation.translate(translator)
 
     def __str__(self) -> str:
-        return  f'{self.duration}\n{self.source_text}\n\n'
-    
+        return f'{self.duration}\n{self.source_text}\n\n'
+
     def get_trans_str(self) -> str:
         return f'{self.duration}\n{self.translation}\n\n'
-    
+
     def get_bilingual_str(self) -> str:
         return f'{self.duration}\n{self.source_text}\n{self.translation}\n\n'
 
+
 class SRT_script():
     def __init__(self, segments) -> None:
         self.segments = []
@@ -80,29 +109,41 @@ class SRT_script():
             self.segments.append(srt_seg)
 
     @classmethod
-    def parse_from_srt_file(cls, path:str):
+    def parse_from_srt_file(cls, path: str):
         with open(path, 'r', encoding="utf-8") as f:
-            script_lines = f.read().splitlines() 
+            script_lines = [line.rstrip() for line in f.readlines()]
 
         segments = []
         for i in range(len(script_lines)):
             if i % 4 == 0:
-                segments.append(list(script_lines[i:i+4]))
+                segments.append(list(script_lines[i:i + 4]))
 
         return cls(segments)
 
     def merge_segs(self, idx_list) -> SRT_segment:
-        final_seg = self.segments[idx_list[0]]
+        """
+        Merge entire segment list to a single segment
+        :param idx_list: List of index to merge
+        :return: Merged list
+        """
+        if not idx_list:
+            raise NotImplementedError('Empty idx_list')
+        seg_result = deepcopy(self.segments[idx_list[0]])
         if len(idx_list) == 1:
-            return final_seg
-        
+            return seg_result
+
         for idx in range(1, len(idx_list)):
-            final_seg.merge_seg(self.segments[idx_list[idx]])
-        
-        return final_seg
+            seg_result += self.segments[idx_list[idx]]
+
+        return seg_result
 
     def form_whole_sentence(self):
-        merge_list = [] # a list of indices that should be merged e.g. [[0], [1, 2, 3, 4], [5, 6], [7]]
+        """
+        Concatenate or Strip sentences and reconstruct segments list. This is because of
+        improper segmentation from openai-whisper.
+        :return: None
+        """
+        merge_list = []  # a list of indices that should be merged e.g. [[0], [1, 2, 3, 4], [5, 6], [7]]
         sentence = []
         for i, seg in enumerate(self.segments):
             if seg.source_text[-1] in ['.', '!', '?'] and len(seg.source_text) > 10 and 'vs.' not in seg.source_text:
@@ -117,113 +158,117 @@ class SRT_script():
             segments.append(self.merge_segs(idx_list))
 
         self.segments = segments
-    
+
     def remove_trans_punctuation(self):
-        # Post-process: remove all punc after translation and split
+        """
+        Post-process: remove all punc after translation and split
+        :return: None
+        """
         for i, seg in enumerate(self.segments):
             seg.remove_trans_punc()
 
-    def set_translation(self, translate:str, id_range:tuple, model, video_name, video_link=None):
+    def set_translation(self, translate: str, id_range: tuple, model, video_name, video_link=None):
         start_seg_id = id_range[0]
         end_seg_id = id_range[1]
-        
+
         src_text = ""
-        for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
-            src_text+=seg.source_text
-            src_text+='\n\n'
+        for i, seg in enumerate(self.segments[start_seg_id - 1:end_seg_id]):
+            src_text += seg.source_text
+            src_text += '\n\n'
 
-        def inner_func(target,input_str):
+        def inner_func(target, input_str):
             response = openai.ChatCompletion.create(
-                    #model=model,
-                    model = "gpt-3.5-turbo",
-                    messages = [
-                        #{"role": "system", "content": "You are a helpful assistant that help calibrates English to Chinese subtitle translations in starcraft2."},
-                        #{"role": "system", "content": "You are provided with a translated Chinese transcript; you must modify or split the Chinese sentence to match the meaning and the number of the English transcript exactly one by one. You must not merge ANY Chinese lines, you can only split them but the total Chinese lines MUST equals to number of English lines."},
-                        #{"role": "system", "content": "There is no need for you to add any comments or notes, and do not modify the English transcript."},
-                        #{"role": "user", "content": 'You are given the English transcript and line number, your task is to merge or split the Chinese to match the exact number of lines in English transcript, no more no less. For example, if there are more Chinese lines than English lines, merge some the Chinese lines to match the number of English lines. If Chinese lines is less than English lines, split some Chinese lines to match the english lines: "{}"'.format(input_str)}
-
-                        {"role": "system", "content": "你的任务是按照要求合并或拆分句子到指定行数，你需要尽可能保证句意，但必要时可以将一句话分为两行输出"},
-                        {"role": "system", "content": "注意：你只需要输出处理过的中文句子，如果你要输出序号，请使用冒号隔开"},
-                        {"role": "user", "content": '请将下面的句子拆分或组合为{}句:\n{}'.format(target,input_str)}
-                        # {"role": "system", "content": "请将以下中文与其英文句子一一对应并输出："},
-                        # {"role": "system", "content": "英文：{}".format(src_text)},
-                        # {"role": "user", "content": "中文：{}\n\n".format(input_str)},
-                    ],
-                    temperature = 0.15
-                )
+                # model=model,
+                model="gpt-3.5-turbo",
+                messages=[
+                    # {"role": "system", "content": "You are a helpful assistant that help calibrates English to Chinese subtitle translations in starcraft2."},
+                    # {"role": "system", "content": "You are provided with a translated Chinese transcript; you must modify or split the Chinese sentence to match the meaning and the number of the English transcript exactly one by one. You must not merge ANY Chinese lines, you can only split them but the total Chinese lines MUST equals to number of English lines."},
+                    # {"role": "system", "content": "There is no need for you to add any comments or notes, and do not modify the English transcript."},
+                    # {"role": "user", "content": 'You are given the English transcript and line number, your task is to merge or split the Chinese to match the exact number of lines in English transcript, no more no less. For example, if there are more Chinese lines than English lines, merge some the Chinese lines to match the number of English lines. If Chinese lines is less than English lines, split some Chinese lines to match the english lines: "{}"'.format(input_str)}
+
+                    {"role": "system",
+                     "content": "你的任务是按照要求合并或拆分句子到指定行数，你需要尽可能保证句意，但必要时可以将一句话分为两行输出"},
+                    {"role": "system", "content": "注意：你只需要输出处理过的中文句子，如果你要输出序号，请使用冒号隔开"},
+                    {"role": "user", "content": '请将下面的句子拆分或组合为{}句:\n{}'.format(target, input_str)}
+                    # {"role": "system", "content": "请将以下中文与其英文句子一一对应并输出："},
+                    # {"role": "system", "content": "英文：{}".format(src_text)},
+                    # {"role": "user", "content": "中文：{}\n\n".format(input_str)},
+                ],
+                temperature=0.15
+            )
             # print(src_text)
             # print(input_str)
             # print(response['choices'][0]['message']['content'].strip())
             # exit()
             return response['choices'][0]['message']['content'].strip()
-        
-        
+
         lines = translate.split('\n\n')
         if len(lines) < (end_seg_id - start_seg_id + 1):
             count = 0
             solved = True
-            while count<5 and len(lines) != (end_seg_id - start_seg_id + 1):
+            while count < 5 and len(lines) != (end_seg_id - start_seg_id + 1):
                 count += 1
                 print("Solving Unmatched Lines|iteration {}".format(count))
-                #input_str = "\n"
-                #initialize GPT input
-                #for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
+                # input_str = "\n"
+                # initialize GPT input
+                # for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
                 #    input_str += 'Sentence %d: ' %(i+1)+ seg.source_text + '\n'
                 #    #Append to prompt string
                 #    #Adds sentence index let GPT keep track of sentence breaks
-                #input_str += translate    
-                #append translate to prompt
+                # input_str += translate
+                # append translate to prompt
                 flag = True
                 while flag:
                     flag = False
-                    #print("translate:")
-                    #print(translate)
+                    # print("translate:")
+                    # print(translate)
                     try:
-                        #print("target")
-                        #print(end_seg_id - start_seg_id + 1)
-                        translate = inner_func(end_seg_id - start_seg_id + 1,translate)
+                        # print("target")
+                        # print(end_seg_id - start_seg_id + 1)
+                        translate = inner_func(end_seg_id - start_seg_id + 1, translate)
                     except Exception as e:
-                        print("An error has occurred during solving unmatched lines:",e)
+                        print("An error has occurred during solving unmatched lines:", e)
                         print("Retrying...")
                         flag = True
                 lines = translate.split('\n')
-                #print("result")
-                #print(len(lines))
-                
+                # print("result")
+                # print(len(lines))
+
             if len(lines) < (end_seg_id - start_seg_id + 1):
                 solved = False
                 print("Failed Solving unmatched lines, Manually parse needed")
-            
+
             if not os.path.exists("./logs"):
                 os.mkdir("./logs")
             if video_link:
                 log_file = "./logs/log_link.csv"
                 log_exist = os.path.exists(log_file)
-                with open(log_file,"a") as log:
+                with open(log_file, "a") as log:
                     if not log_exist:
                         log.write("range_of_text,iterations_solving,solved,file_length,video_link" + "\n")
-                    log.write(str(id_range)+','+str(count)+','+str(solved)+','+str(len(self.segments))+','+video_link + "\n")
+                    log.write(str(id_range) + ',' + str(count) + ',' + str(solved) + ',' + str(
+                        len(self.segments)) + ',' + video_link + "\n")
             else:
                 log_file = "./logs/log_name.csv"
                 log_exist = os.path.exists(log_file)
-                with open(log_file,"a") as log:
+                with open(log_file, "a") as log:
                     if not log_exist:
                         log.write("range_of_text,iterations_solving,solved,file_length,video_name" + "\n")
-                    log.write(str(id_range)+','+str(count)+','+str(solved)+','+str(len(self.segments))+','+video_name + "\n")
-            
+                    log.write(str(id_range) + ',' + str(count) + ',' + str(solved) + ',' + str(
+                        len(self.segments)) + ',' + video_name + "\n")
+
             print(lines)
-            #print(id_range)
-            #for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
+            # print(id_range)
+            # for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
             #    print(seg.source_text)
-            #print(translate)
-            
-        
-        for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
+            # print(translate)
+
+        for i, seg in enumerate(self.segments[start_seg_id - 1:end_seg_id]):
             # naive way to due with merge translation problem
             # TODO: need a smarter solution
 
             if i < len(lines):
-                if "Note:" in lines[i]: # to avoid note
+                if "Note:" in lines[i]:  # to avoid note
                     lines.remove(lines[i])
                     max_num -= 1
                     if i == len(lines) - 1:
@@ -233,7 +278,6 @@ class SRT_script():
                 except:
                     seg.translation = lines[i]
 
-    
     def split_seg(self, seg, text_threshold, time_threshold):
         # evenly split seg to 2 parts and add new seg into self.segments
 
@@ -251,21 +295,24 @@ class SRT_script():
         src_commas = [m.start() for m in re.finditer(',', source_text)]
         trans_commas = [m.start() for m in re.finditer('，', translation)]
         if len(src_commas) != 0:
-            src_split_idx = src_commas[len(src_commas)//2] if len(src_commas) % 2 == 1 else src_commas[len(src_commas)//2 - 1]
+            src_split_idx = src_commas[len(src_commas) // 2] if len(src_commas) % 2 == 1 else src_commas[
+                len(src_commas) // 2 - 1]
         else:
             src_space = [m.start() for m in re.finditer(' ', source_text)]
-            if len(src_space) > 0: 
-                src_split_idx = src_space[len(src_space)//2] if len(src_space) % 2 == 1 else src_space[len(src_space)//2 - 1]
+            if len(src_space) > 0:
+                src_split_idx = src_space[len(src_space) // 2] if len(src_space) % 2 == 1 else src_space[
+                    len(src_space) // 2 - 1]
             else:
                 src_split_idx = 0
 
         if len(trans_commas) != 0:
-            trans_split_idx = trans_commas[len(trans_commas)//2] if len(trans_commas) % 2 == 1 else trans_commas[len(trans_commas)//2 - 1]
+            trans_split_idx = trans_commas[len(trans_commas) // 2] if len(trans_commas) % 2 == 1 else trans_commas[
+                len(trans_commas) // 2 - 1]
         else:
-            trans_split_idx = len(translation)//2
-        
+            trans_split_idx = len(translation) // 2
+
         # split the time duration based on text length
-        time_split_ratio = trans_split_idx/(len(seg.translation) - 1)
+        time_split_ratio = trans_split_idx / (len(seg.translation) - 1)
 
         src_seg1 = source_text[:src_split_idx]
         src_seg2 = source_text[src_split_idx:]
@@ -273,7 +320,7 @@ class SRT_script():
         trans_seg2 = translation[trans_split_idx:]
 
         start_seg1 = seg.start
-        end_seg1 = start_seg2 = seg.start + (seg.end - seg.start)*time_split_ratio
+        end_seg1 = start_seg2 = seg.start + (seg.end - seg.start) * time_split_ratio
         end_seg2 = seg.end
 
         seg1_dict = {}
@@ -295,7 +342,7 @@ class SRT_script():
             result_list += self.split_seg(seg1, text_threshold, time_threshold)
         else:
             result_list.append(seg1)
-        
+
         if len(seg2.translation) > text_threshold and (seg2.end - seg2.start) > time_threshold:
             result_list += self.split_seg(seg2, text_threshold, time_threshold)
         else:
@@ -303,7 +350,6 @@ class SRT_script():
 
         return result_list
 
-
     def check_len_and_split(self, text_threshold=30, time_threshold=1.0):
         # DEPRECATED
         # if sentence length >= threshold and sentence duration > time_threshold, split this segments to two
@@ -314,7 +360,7 @@ class SRT_script():
                 segments += seg_list
             else:
                 segments.append(seg)
-        
+
         self.segments = segments
 
         pass
@@ -325,23 +371,23 @@ class SRT_script():
         end_seg_id = range[1]
         extra_len = 0
         segments = []
-        for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
+        for i, seg in enumerate(self.segments[start_seg_id - 1:end_seg_id]):
             if len(seg.translation) > text_threshold and (seg.end - seg.start) > time_threshold:
                 seg_list = self.split_seg(seg, text_threshold, time_threshold)
                 segments += seg_list
-                extra_len += len(seg_list) - 1 
+                extra_len += len(seg_list) - 1
             else:
                 segments.append(seg)
-        
-        self.segments[start_seg_id-1:end_seg_id] = segments
+
+        self.segments[start_seg_id - 1:end_seg_id] = segments
         return extra_len
 
     def correct_with_force_term(self):
         ## force term correction
 
         # load term dictionary
-        with open("./finetune_data/dict_enzh.csv",'r', encoding='utf-8') as f:
-            term_enzh_dict = {rows[0]:rows[1] for rows in reader(f)}
+        with open("./finetune_data/dict_enzh.csv", 'r', encoding='utf-8') as f:
+            term_enzh_dict = {rows[0]: rows[1] for rows in reader(f)}
 
         # change term
         for seg in self.segments:
@@ -359,7 +405,7 @@ class SRT_script():
 
     def spell_check_term(self):
         ## known bug: I've will be replaced because i've is not in the dict
-        
+
         import enchant
         dict = enchant.Dict('en_US')
         term_spellDict = enchant.PyPWL('./finetune_data/dict_freq.txt')
@@ -371,6 +417,7 @@ class SRT_script():
                 [real_word, pos] = self.get_real_word(word)
                 if not dict.check(word[:pos]):
                     suggest = term_spellDict.suggest(real_word)
+
                     if suggest and enchant.utils.levenshtein(word, suggest[0]) < (len(word)+len(suggest[0]))/4:  # relax spell check
                         
                         #with open("dislog.log","a") as log:
@@ -389,14 +436,13 @@ class SRT_script():
             seg.source_text = " ".join(ready_words)
         pass
 
-    def spell_correction(self, word:str, arg:int):
+    def spell_correction(self, word: str, arg: int):
         try:
-            arg in [0,1]
+            arg in [0, 1]
         except ValueError:
             print('only 0 or 1 for argument')
 
-
-        def uncover(word:str):
+        def uncover(word: str):
             if word[-2:] == ".\n":
                 real_word = word[:-2].lower()
                 n = -2
@@ -406,14 +452,14 @@ class SRT_script():
             else:
                 real_word = word.lower()
                 n = 0
-            return real_word, len(word)+n
-        
+            return real_word, len(word) + n
+
         real_word = uncover(word)[0]
         pos = uncover(word)[1]
         new_word = word
         if arg == 0:  # term translate mode
-            with open("finetune_data/dict_enzh.csv",'r', encoding='utf-8') as f:
-                term_enzh_dict = {rows[0]:rows[1] for rows in reader(f)}
+            with open("finetune_data/dict_enzh.csv", 'r', encoding='utf-8') as f:
+                term_enzh_dict = {rows[0]: rows[1] for rows in reader(f)}
             if real_word in term_enzh_dict:
                 new_word = word.replace(word[:pos], term_enzh_dict.get(real_word))
         elif arg == 1:  # term spell check mode
@@ -422,10 +468,10 @@ class SRT_script():
             term_spellDict = enchant.PyPWL('./finetune_data/dict_freq.txt')
             if not dict.check(real_word):
                 if term_spellDict.suggest(real_word):  # relax spell check
-                    new_word = word.replace(word[:pos],term_spellDict.suggest(real_word)[0])
+                    new_word = word.replace(word[:pos], term_spellDict.suggest(real_word)[0])
         return new_word
-    
-    def get_real_word(self, word:str):
+
+    def get_real_word(self, word: str):
         if word[-2:] == ".\n":
             real_word = word[:-2].lower()
             n = -2
@@ -435,8 +481,7 @@ class SRT_script():
         else:
             real_word = word.lower()
             n = 0
-        return real_word, len(word)+n
-
+        return real_word, len(word) + n
 
     ## WRITE AND READ FUNCTIONS ##
 
@@ -444,48 +489,48 @@ class SRT_script():
         # return a string with pure source text
         result = ""
         for i, seg in enumerate(self.segments):
-            result+=f'SENTENCE {i+1}: {seg.source_text}\n\n\n'
-        
+            result += f'SENTENCE {i + 1}: {seg.source_text}\n\n\n'
+
         return result
-    
+
     def reform_src_str(self):
         result = ""
         for i, seg in enumerate(self.segments):
-            result += f'{i+1}\n'
+            result += f'{i + 1}\n'
             result += str(seg)
         return result
 
     def reform_trans_str(self):
         result = ""
         for i, seg in enumerate(self.segments):
-            result += f'{i+1}\n'
+            result += f'{i + 1}\n'
             result += seg.get_trans_str()
         return result
-    
+
     def form_bilingual_str(self):
         result = ""
         for i, seg in enumerate(self.segments):
-            result += f'{i+1}\n'
+            result += f'{i + 1}\n'
             result += seg.get_bilingual_str()
         return result
 
-    def write_srt_file_src(self, path:str):
+    def write_srt_file_src(self, path: str):
         # write srt file to path
         with open(path, "w", encoding='utf-8') as f:
             f.write(self.reform_src_str())
         pass
 
-    def write_srt_file_translate(self, path:str):
+    def write_srt_file_translate(self, path: str):
         with open(path, "w", encoding='utf-8') as f:
             f.write(self.reform_trans_str())
         pass
 
-    def write_srt_file_bilingual(self, path:str):
+    def write_srt_file_bilingual(self, path: str):
         with open(path, "w", encoding='utf-8') as f:
             f.write(self.form_bilingual_str())
         pass
 
-    def realtime_write_srt(self,path,range,length, idx):
+    def realtime_write_srt(self, path, range, length, idx):
         # DEPRECATED
         start_seg_id = range[0]
         end_seg_id = range[1]
@@ -494,22 +539,20 @@ class SRT_script():
             #     f.write(f'{i+idx}\n')
             #     f.write(seg.get_trans_str())
             for i, seg in enumerate(self.segments):
-                if i<range[0]-1 : continue
-                if i>=range[1] + length :break
-                f.write(f'{i+idx}\n')
+                if i < range[0] - 1: continue
+                if i >= range[1] + length: break
+                f.write(f'{i + idx}\n')
                 f.write(seg.get_trans_str())
         pass
 
-    def realtime_bilingual_write_srt(self,path,range, length,idx):
+    def realtime_bilingual_write_srt(self, path, range, length, idx):
         # DEPRECATED
         start_seg_id = range[0]
         end_seg_id = range[1]
         with open(path, "a", encoding='utf-8') as f:
             for i, seg in enumerate(self.segments):
-                if i<range[0]-1 : continue
-                if i>=range[1] + length :break
-                f.write(f'{i+idx}\n')
+                if i < range[0] - 1: continue
+                if i >= range[1] + length: break
+                f.write(f'{i + idx}\n')
                 f.write(seg.get_bilingual_str())
         pass
-    
-