[fix] #61 - sanitize filenames

dkb_robo/api.py

@@ -11,7 +11,7 @@
 import threading
 from typing import Dict, List, Tuple
 import requests
-from dkb_robo.utilities import get_dateformat
+from dkb_robo.utilities import get_dateformat, get_valid_filename
 from dkb_robo.legacy import Wrapper as Legacywrapper
 
 
@@ -707,7 +707,7 @@ def _download_document(self, path: str, document: Dict[str, str]) -> str:
 
             if response.status_code == 200:
                 self.logger.info('Saving %s/%s...', directories[1], document['filename'])
-                with open(f'{directories[1]}/{document["filename"]}', 'wb') as file:
+                with open(f'{directories[1]}/{get_valid_filename(document["filename"])}', 'wb') as file:
                     file.write(response.content)
 
                 if not document['read']:
@@ -777,7 +777,7 @@ def _process_document(self, path: str, prepend_date: bool, document: Dict[str, s
             documentname_list.append(document['filename'])
 
         self.logger.debug('api.Wrapper._process_document() ended\n')
-        return documentname_list, f'{path}/{document["document_type"]}/{document["filename"]}', rcode
+        return documentname_list, f'{path}/{document["document_type"]}/{get_valid_filename(document["filename"])}', rcode
 
     def _filter_postbox(self, msg_dic: Dict[str, str], pb_dic: Dict[str, str], path: bool = None, download_all: bool = False, _archive: bool = False, prepend_date: bool = None) -> Dict[str, str]:
         """ filter postbox """

dkb_robo/utilities.py

@@ -6,6 +6,7 @@
 from typing import List, Tuple
 from datetime import datetime, timezone
 import time
+import re
 
 
 def get_dateformat():
@@ -45,6 +46,15 @@ def generate_random_string(length: int) -> str:
     return ''.join(random.choice(char_set) for _ in range(length))
 
 
+def get_valid_filename(name):
+    """ sanitize filenames """
+    s = str(name).strip().replace(" ", "_")
+    s = re.sub(r"(?u)[^-\w.]", "_", s)
+    if s in {"", ".", ".."}:
+        s = f'{generate_random_string(8)}.pdf'
+    return s
+
+
 def string2float(value: str) -> float:
     """ convert string to float value """
     try:

test/test_utilities.py

@@ -17,13 +17,14 @@ class TestDKBRobo(unittest.TestCase):
 
     def setUp(self):
         self.dir_path = os.path.dirname(os.path.realpath(__file__))
-        from dkb_robo.utilities import validate_dates, generate_random_string, logger_setup, string2float, _convert_date_format, get_dateformat
+        from dkb_robo.utilities import validate_dates, generate_random_string, logger_setup, string2float, _convert_date_format, get_dateformat, get_valid_filename
         self.validate_dates = validate_dates
         self.string2float = string2float
         self.generate_random_string = generate_random_string
         self.logger_setup = logger_setup
         self._convert_date_format = _convert_date_format
         self.get_dateformat = get_dateformat
+        self.get_valid_filename = get_valid_filename
         self.logger = logging.getLogger('dkb_robo')
 
     @patch('time.time')
@@ -218,7 +219,47 @@ def test_030__convert_date_format(self):
         """ test _convert_date_format() no match """
         self.assertEqual('wrong date', self._convert_date_format(self.logger, 'wrong date', ['%Y/%m/%d', '%Y-%m-%d'], '%d.%m.%Y'))
 
-
+    def test_039__get_valid_filename(self):
+        """ test get_valid_filename """
+        filename = 'test.pdf'
+        self.assertEqual('test.pdf', self.get_valid_filename(filename))
+
+    def test_040__get_valid_filename(self):
+        """ test get_valid_filename """
+        filename = 'test test.pdf'
+        self.assertEqual('test_test.pdf', self.get_valid_filename(filename))
+
+    def test_041__get_valid_filename(self):
+        """ test get_valid_filename """
+        filename = 'testötest.pdf'
+        self.assertEqual('testötest.pdf', self.get_valid_filename(filename))
+
+    def test_042__get_valid_filename(self):
+        """ test get_valid_filename """
+        filename = 'test/test.pdf'
+        self.assertEqual('test_test.pdf', self.get_valid_filename(filename))
+
+    def test_043_get_valid_filename(self):
+        """ test get_valid_filename """
+        filename = 'test\\test.pdf'
+        self.assertEqual('test_test.pdf', self.get_valid_filename(filename))
+
+    def test_044_get_valid_filename(self):
+        """ test get_valid_filename """
+        filename = '.\test.pdf'
+        self.assertEqual('._est.pdf', self.get_valid_filename(filename))
+
+    def test_045_get_valid_filename(self):
+        """ test get_valid_filename """
+        filename = '../test.pdf'
+        self.assertEqual('.._test.pdf', self.get_valid_filename(filename))
+
+    @patch('dkb_robo.utilities.generate_random_string')
+    def test_046_get_valid_filename(self, mock_rand):
+        """ test get_valid_filename """
+        filename = '..'
+        mock_rand.return_value = 'random'
+        self.assertEqual('random.pdf', self.get_valid_filename(filename))
 
 if __name__ == '__main__':