solving conflict

Former-commit-id: f1806577d7055d0cc9b0f3a06cebc11d99a400f9

SRT.py

@@ -109,7 +109,6 @@ class SRT_script():
 
         self.segments = segments # need memory release?
         
-    
 
     def set_translation(self, translate:str, id_range:tuple, model, video_name, video_link=None):
         start_seg_id = id_range[0]
@@ -196,15 +195,18 @@ class SRT_script():
                     seg.translation = lines[i].split("：" or ":")[1]
                 except:
                     seg.translation = lines[i]
-                    #print(lines[i])
-        pass
+
     
     def split_seg(self, seg, threshold):
         # evenly split seg to 2 parts and add new seg into self.segments
-        if seg.source_text[:2] == ', ':
-            seg.source_text = seg.source_text[2:]
+
+        # ignore the initial comma to solve the recursion problem
+        if len(seg.source_text) > 2:
+            if seg.source_text[:2] == ', ':
+                seg.source_text = seg.source_text[2:]
         if seg.translation[0] == '，':
             seg.translation = seg.translation[1:]
+
         source_text = seg.source_text
         translation = seg.translation
         src_commas = [m.start() for m in re.finditer(',', source_text)]

pipeline.py

@@ -133,24 +133,17 @@ else:
 
         # use stable-whisper
         model = stable_whisper.load_model('base')
-        transcript = model.transcribe(audio_path, regroup = False)
-        (
-            transcript
-            .split_by_punctuation(['.', '。', '?'])
-            .merge_by_gap(.15, max_words=3)
-            .merge_by_punctuation([' '])
-            .split_by_punctuation(['.', '。', '?'])
-        )
-        # transcript.to_srt_vtt(srt_file_en)
+        transcript = model.transcribe(audio_path)
+        # (
+        #     transcript
+        #     .split_by_punctuation(['.', '。', '?'])
+        #     .merge_by_gap(.15, max_words=3)
+        #     .merge_by_punctuation([' '])
+        #     .split_by_punctuation(['.', '。', '?'])
+        # )
         transcript = transcript.to_dict()
         srt = SRT_script(transcript['segments']) # read segments to SRT class
 
-        #Write SRT file
-        
-        # from whisper.utils import WriteSRT
-        # with open(srt_file_en, 'w', encoding="utf-8") as f:
-        #     writer = WriteSRT(RESULT_PATH)
-        #     writer.write_result(transcript, f)
     else:
         srt = SRT_script.parse_from_srt_file(srt_file_en)
 
@@ -241,21 +234,6 @@ def get_response(model_name, sentence):
         )
 
         return response['choices'][0]['message']['content'].strip()
-    
-    # if model_name == "text-davinci-003":
-    #     prompt = f"Please help me translate this into Chinese:\n\n{s}\n\n"
-    #     # print(prompt)
-    #     response = openai.Completion.create(
-    #         model=model_name,
-    #         prompt=prompt,
-    #         temperature=0.1,
-    #         max_tokens=2000,
-    #         top_p=1.0,
-    #         frequency_penalty=0.0,
-    #         presence_penalty=0.0
-    #     )
-    #     return response['choices'][0]['text'].strip()
-    pass
         
         
 # Translate and save
@@ -283,7 +261,7 @@ for sentence, range in tqdm(zip(script_arr, range_arr)):
 
 srt.check_len_and_split()
 srt.write_srt_file_translate(f"{RESULT_PATH}/{VIDEO_NAME}/{VIDEO_NAME}_zh.srt")
-# srt.write_srt_file_bilingual(f"{RESULT_PATH}/{VIDEO_NAME}/{VIDEO_NAME}_bi.srt")
+srt.write_srt_file_bilingual(f"{RESULT_PATH}/{VIDEO_NAME}/{VIDEO_NAME}_bi.srt")
 
 if not args.only_srt:
     assSub_zh = srt2ass(f"{RESULT_PATH}/{VIDEO_NAME}/{VIDEO_NAME}_zh.srt", "default", "No", "Modest")