Always use mc:Fallback when reading mc:AlternateContent elements

2 files changed, 21 insertions, 2 deletions

diff --git a/mammoth/docx/office_xml.py b/mammoth/docx/office_xml.py
index 4dffb44..0a6a8a8 100644
--- a/mammoth/docx/office_xml.py
+++ b/mammoth/docx/office_xml.py
@@ -1,4 +1,5 @@
-from .xmlparser import parse_xml
+from ..lists import flat_map
+from .xmlparser import parse_xml, XmlElement
 
 
 _namespaces = [
@@ -16,5 +17,15 @@ _namespaces = [
 
 
 def read(fileobj):
-    return parse_xml(fileobj, _namespaces)
+    return _collapse_alternate_content(parse_xml(fileobj, _namespaces))[0]
     
+
+def _collapse_alternate_content(node):
+    if isinstance(node, XmlElement):
+        if node.name == "mc:AlternateContent":
+            return node.find_child("mc:Fallback").children
+        else:
+            node.children = flat_map(_collapse_alternate_content, node.children)
+            return [node]
+    else:
+        return [node]
diff --git a/mammoth/lists.py b/mammoth/lists.py
index 6e39ae0..e8477cb 100644
--- a/mammoth/lists.py
+++ b/mammoth/lists.py
@@ -24,6 +24,14 @@ def unique(values):
     return output
 
 
+def flat_map(func, values):
+    return [
+        element
+        for value in values
+        for element in func(value)
+    ]
+
+
 if sys.version_info[0] == 2:
     map = map
     filter = filter