antimoda1 commited on
Commit
e43669b
·
1 Parent(s): 7a668f2
Files changed (1) hide show
  1. _1_get_documents.py +1 -2
_1_get_documents.py CHANGED
@@ -13,7 +13,7 @@ def process_file(file_path):
13
  with open(file_path, 'r', encoding='utf-8-sig') as f:
14
  text = f.read()
15
  assert text
16
- return str(file_path).split('.')[-1], text
17
 
18
  def process_folder_recursive(folder_path):
19
  all_messages = []
@@ -28,5 +28,4 @@ def process_folder_recursive(folder_path):
28
  def load_and_process_data() -> list[dict]:
29
  """Загрузка и предобработка данных из JSON файлов"""
30
  all_messages = process_folder_recursive('texts')
31
-
32
  return [x[0] for x in all_messages], [x[1][:-3] for x in all_messages] # возвращаем расширения и тексты документов
 
13
  with open(file_path, 'r', encoding='utf-8-sig') as f:
14
  text = f.read()
15
  assert text
16
+ return text, str(file_path).split('.')[-1]
17
 
18
  def process_folder_recursive(folder_path):
19
  all_messages = []
 
28
  def load_and_process_data() -> list[dict]:
29
  """Загрузка и предобработка данных из JSON файлов"""
30
  all_messages = process_folder_recursive('texts')
 
31
  return [x[0] for x in all_messages], [x[1][:-3] for x in all_messages] # возвращаем расширения и тексты документов