add pre/post process config

Former-commit-id: 3538f88e87cedf5a1ff995b4dc73754b9951fe4d

configs/task_config.yaml

@@ -8,4 +8,10 @@ output_type:
 source_lang: EN
 target_lang: ZH
 field: SC2
-chunk_size: 1000
+chunk_size: 1000
+pre_process: 
+  ON: True
+  spell_check: False
+  term_correct: True
+post_process: 
+  ON: True

src/task.py

@@ -79,6 +79,8 @@ class Task:
         self.target_lang = task_cfg["target_lang"]
         self.source_lang = task_cfg["source_lang"]
         self.field = task_cfg["field"]
+        self.pre_setting = task_cfg["pre_process"]
+        self.post_setting = task_cfg["post_process"]
         self.task_id = task_id
         self.audio_path = None
         self.SRT_Script = None
@@ -155,8 +157,10 @@ class Task:
         self.status = TaskStatus.PRE_PROCESSING
         logging.info("--------------------Start Preprocessing SRT class--------------------")
         self.SRT_Script.form_whole_sentence()
-        # self.SRT_Script.spell_check_term()
-        self.SRT_Script.correct_with_force_term()
+        if self.pre_setting["spell_check"]:
+            self.SRT_Script.spell_check_term()
+        if self.pre_setting["term_correct"]:
+            self.SRT_Script.correct_with_force_term()
         processed_srt_path_src = str(Path(self.task_local_dir) / f'{self.task_id}_processed.srt')
         self.SRT_Script.write_srt_file_src(processed_srt_path_src)
 
@@ -221,9 +225,11 @@ class Task:
     
     def run_pipeline(self):
         self.get_srt_class()
-        self.preprocess()
+        if self.pre_setting["ON"]:
+            self.preprocess()
         self.translation()
-        self.postprocess()
+        if self.post_setting["ON"]:
+            self.postprocess()
         self.result = self.output_render()
         print(self.result)