Read default content types

1 files changed, 21 insertions, 0 deletions

diff --git a/mammoth/docx/content_types_xml.py b/mammoth/docx/content_types_xml.py
new file mode 100644
index 0000000..a2da8e0
--- /dev/null
+++ b/mammoth/docx/content_types_xml.py
@@ -0,0 +1,21 @@
+def read_content_types_xml_element(element):
+    extension_defaults = dict(map(
+        _read_default,
+        element.find_children("content-types:Default"))
+    )
+    return ContentTypes(extension_defaults)
+
+
+def _read_default(element):
+    extension = element.attributes["Extension"]
+    content_type = element.attributes["ContentType"]
+    return extension, content_type
+
+
+class ContentTypes(object):
+    def __init__(self, extension_defaults):
+        self._extension_defaults = extension_defaults
+    
+    def find_content_type(self, path):
+        extension = path.rpartition(".")[2]
+        return self._extension_defaults.get(extension)