[feat] add checksum validation for downloaded files

dkb_robo/postbox.py

@@ -1,5 +1,6 @@
 """ Module for handling the DKB postbox. """
 import datetime
+import hashlib
 import logging
 from dataclasses import dataclass
 from pathlib import Path
@@ -85,6 +86,15 @@ def download(self, client: requests.Session, target_file: Path, overwrite: bool
 
             with target_file.open('wb') as file:
                 file.write(resp.content)
+
+            # compare checksums of file with checksum from document metadata
+            if self.document.checksum:
+                with target_file.open('rb') as file:
+                    checksum = hashlib.md5(file.read()).hexdigest()
+                    if checksum != self.document.checksum:
+                        logger.warning("Checksum mismatch for %s: %s != %s. Renaming file.", target_file, checksum, self.document.checksum)
+                        # rename file to indicate checksum mismatch
+                        target_file.rename(target_file.with_name(target_file.name + '.checksum_mismatch'))
             return True
         return False
 

test/test_postbox.py

@@ -8,6 +8,7 @@
 from dkb_robo.api import DKBRoboError
 from dkb_robo.postbox import PostboxItem, PostBox, Document, Message
 
+
 class TestPostboxItem(unittest.TestCase):
     """ Tests for the PostboxItem class. """
 
@@ -17,7 +18,7 @@ def setUp(self):
             expirationDate="2023-12-31",
             retentionPeriod="9999-12-31",
             contentType="application/pdf",
-            checksum="abc123",
+            checksum="9473fdd0d880a43c21b7778d34872157",
             fileName="test_document",
             metadata={"statementDate": "2023-01-01"},
             owner="owner",
@@ -82,6 +83,20 @@ def test_download_existing_file(self, mock_session):
         self.assertFalse(result)
         target_file.unlink()
 
+    @patch("requests.Session")
+    def test_download_checksum_mismatch(self, mock_session):
+        """ Test that the download method renames the downloaded file if the checksum of the downloaded file does not match."""
+        mock_client = mock_session.return_value
+        mock_client.get.return_value.status_code = 200
+        mock_client.get.return_value.content = b"wrong test content"
+        target_file = Path(tempfile.gettempdir()) / "test_document.pdf"
+        result = self.postbox_item.download(mock_client, target_file, overwrite=True)
+        self.assertTrue(result)
+        self.assertFalse(target_file.exists())
+        mismatched_file = target_file.with_name(target_file.name + ".checksum_mismatch")
+        self.assertTrue(mismatched_file.exists())
+        mismatched_file.unlink()  # Remove the downloaded test file
+
     def test_filename(self):
         """ Test that the filename method returns the correct filename for the postbox item."""
         filename = self.postbox_item.filename()
@@ -185,6 +200,7 @@ def test_date_invalid(self):
         with self.assertRaises(AttributeError):
             self.postbox_item.date()
 
+
 class TestPostBox(unittest.TestCase):
     """ Tests for the PostBox class. """