Convert SDT checkboxes to checkbox inputs

2 files changed, 19 insertions, 2 deletions

diff --git a/mammoth/docx/body_xml.py b/mammoth/docx/body_xml.py
index 134917f..cc4e9ad 100644
--- a/mammoth/docx/body_xml.py
+++ b/mammoth/docx/body_xml.py
@@ -137,7 +137,10 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f
         if element is None:
             return False
         else:
-            return element.attributes.get("w:val") not in ["false", "0"]
+            return read_boolean_attribute_value(element.attributes.get("w:val"))
+
+    def read_boolean_attribute_value(value):
+        return value not in ["false", "0"]
 
     def read_underline_element(element):
         return element and element.attributes.get("w:val") not in [None, "false", "0", "none"]
@@ -569,7 +572,17 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f
         return read_child_elements(element.find_child("mc:Fallback"))
 
     def read_sdt(element):
-        return read_child_elements(element.find_child_or_null("w:sdtContent"))
+        checkbox = element.find_child_or_null("w:sdtPr").find_child("wordml:checkbox")
+
+        if checkbox is not None:
+            checked_element = checkbox.find_child("wordml:checked")
+            is_checked = (
+                checked_element is not None and
+                read_boolean_attribute_value(checked_element.attributes.get("wordml:val"))
+            )
+            return _success(documents.checkbox(checked=is_checked))
+        else:
+            return read_child_elements(element.find_child_or_null("w:sdtContent"))
 
     handlers = {
         "w:t": text,
diff --git a/mammoth/docx/office_xml.py b/mammoth/docx/office_xml.py
index bebd04c..a988894 100644
--- a/mammoth/docx/office_xml.py
+++ b/mammoth/docx/office_xml.py
@@ -23,6 +23,10 @@ _namespaces = [
     ("mc", "http://schemas.openxmlformats.org/markup-compatibility/2006"),
     ("v", "urn:schemas-microsoft-com:vml"),
     ("office-word", "urn:schemas-microsoft-com:office:word"),
+
+    # [MS-DOCX]: Word Extensions to the Office Open XML (.docx) File Format
+    # https://learn.microsoft.com/en-us/openspecs/office_standards/ms-docx/b839fe1f-e1ca-4fa6-8c26-5954d0abbccd
+    ("wordml", "http://schemas.microsoft.com/office/word/2010/wordml"),
 ]