fix real-time output

Former-commit-id: 1fea3d98f33d848dc57260e285ecf9f434c14690

SRT.py

@@ -247,21 +247,23 @@ class SRT_script():
 
         pass
 
-    def check_len_and_split_range(self, range, threshold=30000):
+    def check_len_and_split_range(self, range, threshold=30):
         # TODO: if sentence length >= threshold, split this segments to two
         start_seg_id = range[0]
         end_seg_id = range[1]
+        extra_len = 0
         segments = []
         for i, seg in enumerate(self.segments[start_seg_id-1:end_seg_id]):
             if len(seg.translation) > threshold:
                 seg_list = self.split_seg(seg, threshold)
                 segments += seg_list
+                extra_len += len(seg_list) - 1 
             else:
                 segments.append(seg)
         
         self.segments[start_seg_id-1:end_seg_id] = segments
 
-        return len(segments)
+        return extra_len
     
     def get_source_only(self):
         # return a string with pure source text
@@ -411,7 +413,7 @@ class SRT_script():
             #     f.write(seg.get_trans_str())
             for i, seg in enumerate(self.segments):
                 if i<range[0]-1: continue
-                if i>=range[1]:break
+                if i>=range[1] + length:break
                 f.write(f'{i+idx}\n')
                 f.write(seg.get_trans_str())
         pass
@@ -422,7 +424,7 @@ class SRT_script():
         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]:break
+                if i>=range[1] + length:break
                 f.write(f'{i+idx}\n')
                 f.write(seg.get_bilingual_str())
         pass

pipeline.py

@@ -47,8 +47,8 @@ if args.video_name == 'placeholder' :
         VIDEO_NAME = args.audio_file.split('/')[-1].split('.')[0]
     elif args.srt_file is not None:
         VIDEO_NAME = args.srt_file.split('/')[-1].split('.')[0]
-else:
-    VIDEO_NAME = args.video_name
+    else:
+        VIDEO_NAME = args.video_name
 
 model_name = args.model_name
 
@@ -262,6 +262,7 @@ for sentence, range in tqdm(zip(script_arr, range_arr)):
     print(translate)
     srt.set_translation(translate, range, model_name)
     add_length = srt.check_len_and_split_range(range)
+    print(add_length)
     srt.realtime_write_srt(f"{RESULT_PATH}/{VIDEO_NAME}/{VIDEO_NAME}_zh.srt",range, add_length,segidx)
     srt.realtime_bilingual_write_srt(f"{RESULT_PATH}/{VIDEO_NAME}/{VIDEO_NAME}_bi.srt",range, add_length,segidx)