Merge branch 'eason/refactor' of https://github.com/project-kxkg/project-t into eason/refactor

SRT.py

@@ -1,6 +1,8 @@
 from datetime import timedelta
 import os
 import whisper
+from csv import reader
+import re
 
 class SRT_segment(object):
     def __init__(self, *args) -> None:
@@ -154,9 +156,35 @@ class SRT_script():
             f.write(self.form_bilingual_str())
         pass
 
-    def correct_with_force_term():
-        # force term correction
-
+    def correct_with_force_term(self):
+        ## force term correction
+        # TODO: shortcut translation i.e. VA, ob
+        # TODO: variety of translation
+
+        # load term dictionary
+        with open("finetune_data/dict.csv",'r', encoding='utf-8') as f:
+            csv_reader = reader(f)
+            term_dict = {rows[0]:rows[1] for rows in csv_reader}
+
+        # change term
+        for seg in self.segments:
+            ready_words = re.sub('\n', '\n ', seg.source_text).split(" ")
+            for i in range(len(ready_words)):
+                word = ready_words[i]
+                if word[-2:] == ".\n" :
+                    if word[:-2].lower() in term_dict :
+                        new_word = word.replace(word[:-2], term_dict.get(word[:-2].lower())) + ' '
+                        ready_words[i] = new_word
+                    else:
+                        ready_words[i] = word + ' '
+                elif word.lower() in term_dict :
+                    new_word = word.replace(word,term_dict.get(word.lower())) + ' '
+                    ready_words[i] = new_word
+                else :
+                    ready_words[i]= word + ' '
+            seg.source_text = re.sub('\n ', '\n', "".join(ready_words))
+        
+        print(self)
         pass