4 files changed, 29 insertions, 12 deletions

diff --git a/libmat2/office.py b/libmat2/office.py
index 5165056..6087c47 100644
--- a/libmat2/office.py
+++ b/libmat2/office.py
@@ -147,7 +147,10 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
         """ In this function, we're changing the XML
         document in two times, since we don't want
         to change the tree we're iterating on."""
-        tree, ns = _parse_xml(full_path)
+        try:
+            tree, ns = _parse_xml(full_path)
+        except ET.ParseError:
+            return False
 
         # No revisions are present
         del_presence = tree.find('.//w:del', ns)
@@ -191,15 +194,13 @@ class MSOfficeParser(ArchiveBasedAbstractParser):
         zipin = zipfile.ZipFile(self.filename)
         for item in zipin.infolist():
             if item.filename.startswith('docProps/') and item.filename.endswith('.xml'):
-                content = zipin.read(item).decode('utf-8')
                 try:
+                    content = zipin.read(item).decode('utf-8')
                     results = re.findall(r"<(.+)>(.+)</\1>", content, re.I|re.M)
                     for (key, value) in results:
                         metadata[key] = value
-                except TypeError:  # We didn't manage to parse the xml file
-                    pass
-                if not metadata:  # better safe than sorry
-                    metadata[item] = 'harmful content'
+                except (TypeError, UnicodeDecodeError):  # We didn't manage to parse the xml file
+                    metadata[item.filename] = 'harmful content'
             for key, value in self._get_zipinfo_meta(item).items():
                 metadata[key] = value
         zipin.close()
@@ -232,7 +233,10 @@ class LibreOfficeParser(ArchiveBasedAbstractParser):
 
 
     def __remove_revisions(self, full_path: str) -> bool:
-        tree, ns = _parse_xml(full_path)
+        try:
+            tree, ns = _parse_xml(full_path)
+        except ET.ParseError:
+            return False
 
         if 'office' not in ns.keys():  # no revisions in the current file
             return True
@@ -259,15 +263,13 @@ class LibreOfficeParser(ArchiveBasedAbstractParser):
         zipin = zipfile.ZipFile(self.filename)
         for item in zipin.infolist():
             if item.filename == 'meta.xml':
-                content = zipin.read(item).decode('utf-8')
                 try:
+                    content = zipin.read(item).decode('utf-8')
                     results = re.findall(r"<((?:meta|dc|cp).+?)>(.+)</\1>", content, re.I|re.M)
                     for (key, value) in results:
                         metadata[key] = value
-                except TypeError:  # We didn't manage to parse the xml file
-                    pass
-                if not metadata:  # better safe than sorry
-                    metadata[item] = 'harmful content'
+                except (TypeError, UnicodeDecodeError):  # We didn't manage to parse the xml file
+                    metadata[item.filename] = 'harmful content'
             for key, value in self._get_zipinfo_meta(item).items():
                 metadata[key] = value
         zipin.close()
diff --git a/tests/data/embedded_corrupted.docx b/tests/data/embedded_corrupted.docx
new file mode 100644
index 0000000..989bdb8
--- /dev/null
+++ b/tests/data/embedded_corrupted.docx
diff --git a/tests/data/embedded_corrupted.odt b/tests/data/embedded_corrupted.odt
new file mode 100644
index 0000000..1e4a844
--- /dev/null
+++ b/tests/data/embedded_corrupted.odt
diff --git a/tests/test_corrupted_files.py b/tests/test_corrupted_files.py
index a77acbc..2bb1c76 100644
--- a/tests/test_corrupted_files.py
+++ b/tests/test_corrupted_files.py
@@ -15,6 +15,21 @@ class TestUnsupportedFiles(unittest.TestCase):
         self.assertEqual(parser, None)
         os.remove('./tests/clean.py')
 
+class TestCorruptedEmbedded(unittest.TestCase):
+    def test_docx(self):
+        shutil.copy('./tests/data/embedded_corrupted.docx', './tests/data/clean.docx')
+        parser, mimetype = parser_factory.get_parser('./tests/data/clean.docx')
+        self.assertFalse(parser.remove_all())
+        self.assertIsNotNone(parser.get_meta())
+        os.remove('./tests/data/clean.docx')
+
+    def test_odt(self):
+        shutil.copy('./tests/data/embedded_corrupted.odt', './tests/data/clean.odt')
+        parser, mimetype = parser_factory.get_parser('./tests/data/clean.odt')
+        self.assertFalse(parser.remove_all())
+        self.assertEqual(parser.get_meta(), {'create_system': 'Weird', 'date_time': '2018-06-10 17:18:18', 'meta.xml': 'harmful content'})
+        os.remove('./tests/data/clean.odt')
+
 
 class TestExplicitelyUnsupportedFiles(unittest.TestCase):
     def test_pdf(self):