DOCX Input: Fix figures in newer Word documents being duplicated. Fixes #1789238 [textboxes duplcated after import from docx](https://bugs.launchpad.net/calibre/+bug/1789238)

Newer Word versions use a proprietary image markup with a fallback to
the standard form. Ignore the proprietary form and use only the
fallback, standard version when converting.

src/calibre/ebooks/docx/names.py

@@ -39,6 +39,7 @@ TRANSITIONAL_NAMESPACES = {
     'mo': 'http://schemas.microsoft.com/office/mac/office/2008/main',
     'o': 'urn:schemas-microsoft-com:office:office',
     've': 'http://schemas.openxmlformats.org/markup-compatibility/2006',
+    'mc': 'http://schemas.openxmlformats.org/markup-compatibility/2006',
     # Text Content
     'w': 'http://schemas.openxmlformats.org/wordprocessingml/2006/main',
     'w10': 'urn:schemas-microsoft-com:office:word',

src/calibre/ebooks/docx/to_html.py

@@ -120,6 +120,7 @@ class Convert(object):
         self.log.debug('Converting Word markup to HTML')
 
         self.read_page_properties(doc)
+        self.resolve_alternate_content(doc)
         self.current_rels = relationships_by_id
         for wp, page_properties in self.page_map.iteritems():
             self.current_page = page_properties
@@ -267,6 +268,17 @@ class Convert(object):
             for x in current:
                 self.page_map[x] = pr
 
+    def resolve_alternate_content(self, doc):
+        # For proprietary extensions in Word documents use the fallback, spec
+        # compliant form
+        # See https://wiki.openoffice.org/wiki/OOXML/Markup_Compatibility_and_Extensibility
+        for ac in self.namespace.descendants(doc, 'mc:AlternateContent'):
+            choices = self.namespace.XPath('./mc:Choice')(ac)
+            fallbacks = self.namespace.XPath('./mc:Fallback')(ac)
+            if fallbacks:
+                for choice in choices:
+                    ac.remove(choice)
+
     def read_styles(self, relationships_by_type):
 
         def get_name(rtype, defname):