panchadip commited on
Commit
46cc02c
·
verified ·
1 Parent(s): a70fc0e

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +15 -6
app.py CHANGED
@@ -9,16 +9,25 @@ st.sidebar.title("Whatsapp Chat Analyzer")
9
  uploaded_file = st.sidebar.file_uploader("Choose a file")
10
  if uploaded_file is not None:
11
  bytes_data = uploaded_file.read()
12
-
13
- # Detect encoding
14
  result = chardet.detect(bytes_data)
15
  encoding = result['encoding']
16
-
17
- # Decode the bytes to string using the detected encoding
18
- data = bytes_data.decode(encoding)
 
 
 
 
 
 
 
 
 
19
  df = preprocessor.preprocess(data)
20
 
21
- # fetch unique users
22
  user_list = df['user'].unique().tolist()
23
  user_list.remove('group_notification')
24
  user_list.sort()
 
9
  uploaded_file = st.sidebar.file_uploader("Choose a file")
10
  if uploaded_file is not None:
11
  bytes_data = uploaded_file.read()
12
+
13
+ # Detect the encoding
14
  result = chardet.detect(bytes_data)
15
  encoding = result['encoding']
16
+
17
+ # Fallback to utf-8 if encoding is None
18
+ if encoding is None:
19
+ encoding = 'utf-8' # or 'latin1' if utf-8 fails
20
+
21
+ # Decode with the detected or fallback encoding
22
+ try:
23
+ data = bytes_data.decode(encoding)
24
+ except UnicodeDecodeError:
25
+ # If utf-8 decoding fails, fallback to latin1
26
+ data = bytes_data.decode('latin1')
27
+
28
  df = preprocessor.preprocess(data)
29
 
30
+ # Fetch unique users
31
  user_list = df['user'].unique().tolist()
32
  user_list.remove('group_notification')
33
  user_list.sort()