Strip comment references from document.xml

1 files changed, 31 insertions, 1 deletions

diff --git a/tests/test_libmat2.py b/tests/test_libmat2.py
index 0435113..491f396 100644
--- a/tests/test_libmat2.py
+++ b/tests/test_libmat2.py
@@ -900,4 +900,34 @@ class TextDocx(unittest.TestCase):
             self.assertIsNotNone(match)
 
         os.remove('./tests/data/comment_clean.docx')
-        os.remove('./tests/data/comment_clean.cleaned.docx')
\ No newline at end of file
+        os.remove('./tests/data/comment_clean.cleaned.docx')
+
+    def test_comment_references_are_removed(self):
+        with zipfile.ZipFile('./tests/data/comment.docx') as zipin:
+            c = zipin.open('word/document.xml')
+            content = c.read()
+
+            r = b'w:commentRangeStart'
+            self.assertIn(r, content)
+            r = b'w:commentRangeEnd'
+            self.assertIn(r, content)
+            r = b'w:commentReference'
+            self.assertIn(r, content)
+
+        shutil.copy('./tests/data/comment.docx', './tests/data/comment_clean.docx')
+        p = office.MSOfficeParser('./tests/data/comment_clean.docx')
+        self.assertTrue(p.remove_all())
+
+        with zipfile.ZipFile('./tests/data/comment_clean.cleaned.docx') as zipin:
+            c = zipin.open('word/document.xml')
+            content = c.read()
+
+            r = b'w:commentRangeStart'
+            self.assertNotIn(r, content)
+            r = b'w:commentRangeEnd'
+            self.assertNotIn(r, content)
+            r = b'w:commentReference'
+            self.assertNotIn(r, content)
+
+        os.remove('./tests/data/comment_clean.docx')
+        os.remove('./tests/data/comment_clean.cleaned.docx')