cardinal-pythonlib 2.1.2__tar.gz → 2.1.3__tar.gz

{cardinal_pythonlib-2.1.2 → cardinal_pythonlib-2.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cardinal_pythonlib
-Version: 2.1.2
+Version: 2.1.3
 Summary: Miscellaneous Python libraries
 Home-page: https://github.com/RudolfCardinal/pythonlib
 Author: Rudolf Cardinal

{cardinal_pythonlib-2.1.2 → cardinal_pythonlib-2.1.3}/cardinal_pythonlib/extract_text.py

@@ -1369,13 +1369,14 @@ def _gen_msg_content(
         )
 
     for attachment in message.attachments:
-        # null termination seen in the real world
-        # https://github.com/TeamMsgExtractor/msg-extractor/issues/464
-        ext = attachment.extension.replace("\x00", "")
-        if ext is not None and ext in ext_map:
-            yield document_to_text(
-                blob=attachment.data, extension=ext, config=config
-            )
+        if (extension := getattr(attachment, "extension", None)) is not None:
+            # null termination seen in the real world
+            # https://github.com/TeamMsgExtractor/msg-extractor/issues/464
+            extension = extension.replace("\x00", "")
+            if extension in ext_map:
+                yield document_to_text(
+                    blob=attachment.data, extension=extension, config=config
+                )
 
 
 # =============================================================================

{cardinal_pythonlib-2.1.2 → cardinal_pythonlib-2.1.3}/cardinal_pythonlib/tests/extract_text_tests.py

@@ -33,6 +33,7 @@ import subprocess
 from tempfile import mkdtemp, NamedTemporaryFile
 from unittest import mock, TestCase
 
+from extract_msg import SignedAttachment
 from faker import Faker
 from faker_file.providers.docx_file import DocxFileProvider
 from faker_file.providers.eml_file import EmlFileProvider
@@ -735,3 +736,51 @@ class ConvertMsgToTextTests(ExtractTextTestCase):
             converted = convert_msg_to_text(dummy_filename, config=self.config)
 
         self.assertEqual(converted.strip(), content)
+
+    def test_attachment_with_null_extension_skipped(self) -> None:
+        self.fake.add_provider(DocxFileProvider)
+
+        dummy_filename = "dummy_filename.msg"
+
+        content = self.fake.paragraph(nb_sentences=10)
+        docx = self.fake.docx_file(content=content, raw=True)
+        mock_attachment = mock.Mock(
+            extension=None,
+            data=BytesIO(docx).read(),
+        )
+        mock_msgfile = mock.Mock(
+            body=None, htmlBody=None, attachments=[mock_attachment]
+        )
+        mock_openmsg = mock.Mock(return_value=mock_msgfile)
+        with mock.patch.multiple(
+            "cardinal_pythonlib.extract_text",
+            openMsg=mock_openmsg,
+        ):
+            self.config.width = 0
+            converted = convert_msg_to_text(dummy_filename, config=self.config)
+
+        self.assertEqual(converted.strip(), "")
+
+    def test_signed_attachment_with_no_extension_skipped(self) -> None:
+        self.fake.add_provider(DocxFileProvider)
+
+        dummy_filename = "dummy_filename.msg"
+
+        content = self.fake.paragraph(nb_sentences=10)
+        docx = self.fake.docx_file(content=content, raw=True)
+        mock_attachment = mock.Mock(
+            spec=SignedAttachment,
+            data=BytesIO(docx).read(),
+        )
+        mock_msgfile = mock.Mock(
+            body=None, htmlBody=None, attachments=[mock_attachment]
+        )
+        mock_openmsg = mock.Mock(return_value=mock_msgfile)
+        with mock.patch.multiple(
+            "cardinal_pythonlib.extract_text",
+            openMsg=mock_openmsg,
+        ):
+            self.config.width = 0
+            converted = convert_msg_to_text(dummy_filename, config=self.config)
+
+        self.assertEqual(converted.strip(), "")

{cardinal_pythonlib-2.1.2 → cardinal_pythonlib-2.1.3}/cardinal_pythonlib/version_string.py

@@ -31,5 +31,5 @@ For changelog, see changelog.rst
 
 """
 
-VERSION_STRING = "2.1.2"
+VERSION_STRING = "2.1.3"
 # Use semantic versioning: https://semver.org/

{cardinal_pythonlib-2.1.2 → cardinal_pythonlib-2.1.3}/cardinal_pythonlib.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cardinal-pythonlib
-Version: 2.1.2
+Version: 2.1.3
 Summary: Miscellaneous Python libraries
 Home-page: https://github.com/RudolfCardinal/pythonlib
 Author: Rudolf Cardinal