[fix] Handle importing UTF-16 encoded CSV files #550

This change ensures proper handling of both UTF-8 and UTF-16 encoded CSV files.

Fixes #550

---------

Co-authored-by: Federico Capoano <[email protected]>
(cherry picked from commit 7c7fc4a)

openwisp_radius/base/models.py

@@ -47,6 +47,7 @@
 )
 from ..utils import (
     SmsMessage,
+    decode_byte_data,
     find_available_username,
     generate_sms_token,
     get_sms_default_valid_until,
@@ -951,7 +952,7 @@ def csvfile_upload(
         if not csvfile:
             csvfile = self.csvfile
         csv_data = csvfile.read()
-        csv_data = csv_data.decode('utf-8') if isinstance(csv_data, bytes) else csv_data
+        csv_data = decode_byte_data(csv_data)
         reader = csv.reader(StringIO(csv_data), delimiter=',')
         self.full_clean()
         self.save()

openwisp_radius/receivers.py

@@ -1,6 +1,7 @@
 """
 Receiver functions for django signals (eg: post_save)
 """
+
 import logging
 
 from celery.exceptions import OperationalError

openwisp_radius/tests/static/test_batch_utf8Sig_file2.csv

@@ -0,0 +1 @@
+44D1FADD7379,cleartext$D0weL6L8,[email protected],EAPUSER1,USER1

openwisp_radius/tests/test_utils.py

@@ -24,6 +24,19 @@ def test_validate_file_format(self):
             in error.exception.message
         )
 
+    def test_validate_utf16_file_format(self):
+        utf_16_file_1_format_path = self._get_path('static/test_batch_utf16_file1.csv')
+        assert validate_csvfile(open(utf_16_file_1_format_path, 'rb')) is None
+
+        utf_16_file_2_format_path = self._get_path('static/test_batch_utf16_file2.csv')
+        assert validate_csvfile(open(utf_16_file_2_format_path, 'rb')) is None
+
+    def test_validate_utf8Sig_file_format(self):
+        utf_16_file_2_format_path = self._get_path(
+            'static/test_batch_utf8Sig_file2.csv'
+        )
+        assert validate_csvfile(open(utf_16_file_2_format_path, 'rb')) is None
+
     def test_validate_csvfile(self):
         invalid_csv_path = self._get_path('static/test_batch_invalid.csv')
         improper_csv_path = self._get_path('static/test_batch_improper.csv')

openwisp_radius/utils.py

@@ -129,17 +129,40 @@ def find_available_username(username, users_list, prefix=False):
     return tmp
 
 
+def get_encoding_format(byte_data):
+    # Explicitly handle some common encodings, including utf-16le
+    common_encodings = ['utf-8-sig', 'utf-16', 'utf-16be', 'utf-16le', 'ascii']
+
+    for enc in common_encodings:
+        try:
+            byte_data.decode(enc)
+            return enc
+        except (UnicodeDecodeError, TypeError):
+            continue
+
+    return 'utf-8'
+
+
+def decode_byte_data(data):
+    if isinstance(data, bytes):
+        data = data.decode(get_encoding_format(data))
+        data = data.replace('\x00', '')  # Removing null bytes
+    return data
+
+
 def validate_csvfile(csvfile):
     csv_data = csvfile.read()
+
     try:
-        csv_data = csv_data.decode('utf-8') if isinstance(csv_data, bytes) else csv_data
+        csv_data = decode_byte_data(csv_data)
     except UnicodeDecodeError:
         raise ValidationError(
             _(
                 'Unrecognized file format, the supplied file '
                 'does not look like a CSV file.'
             )
         )
+
     reader = csv.reader(StringIO(csv_data), delimiter=',')
     error_message = 'The CSV contains a line with invalid data,\
                     line number {} triggered the following error: {}'