Read v:imagedata with r:id attribute

3 files changed, 41 insertions, 1 deletions

diff --git a/NEWS b/NEWS
index 6ed8c65..1765b3a 100644
--- a/NEWS
+++ b/NEWS
@@ -2,6 +2,8 @@
 
 * Remove duplicate messages from results.
 
+* Read v:imagedata with r:id attribute.
+
 * Read children of v:roundrect.
 
 * Ignore office-word:wrap, v:shadow and v:shapetype.
diff --git a/mammoth/docx/body_xml.py b/mammoth/docx/body_xml.py
index 42c79a2..23ae944 100644
--- a/mammoth/docx/body_xml.py
+++ b/mammoth/docx/body_xml.py
@@ -199,8 +199,11 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f
         return _combine_results(map(lambda blip: _read_blip(blip, alt_text), blips))
     
     def _read_blip(element, alt_text):
+        return _read_image(lambda: _read_blip_image(element), alt_text)
+    
+    def _read_image(find_image, alt_text):
         try:
-            image_path, open_image = _read_blip_image(element)
+            image_path, open_image = find_image()
         except InvalidFileReferenceError as error:
             return _empty_result_with_messages([results.warning(str(error))])
         
@@ -246,6 +249,10 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f
         
         return image_path, open_image
     
+    def read_imagedata(element):
+        title = element.attributes.get("o:title")
+        return _read_image(lambda: _read_embedded_blip_image(element.attributes["r:id"]), title)
+    
     def note_reference_reader(note_type):
         def note_reference(element):
             return _success(documents.note_reference(note_type, element.attributes["w:id"]))
@@ -276,6 +283,7 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f
         "w:br": br,
         "wp:inline": inline,
         "wp:anchor": inline,
+        "v:imagedata": read_imagedata,
         "w:footnoteReference": note_reference_reader("footnote"),
         "w:endnoteReference": note_reference_reader("endnote"),
         "mc:AlternateContent": alternate_content,
diff --git a/tests/docx/body_xml_tests.py b/tests/docx/body_xml_tests.py
index da662d8..d0288bd 100644
--- a/tests/docx/body_xml_tests.py
+++ b/tests/docx/body_xml_tests.py
@@ -297,6 +297,36 @@ class ReadXmlElementTests(object):
         
     @istest
     @funk.with_context
+    def can_read_imagedata_elements_with_rid_attribute(self, context):
+        imagedata_element = xml_element("v:imagedata", {"r:id": "rId5", "o:title": "It's a hat"})
+        
+        image_bytes = b"Not an image at all!"
+        
+        relationships = Relationships({
+            "rId5": Relationship(target="media/hat.png")
+        })
+        
+        docx_file = context.mock()
+        funk.allows(docx_file).open("word/media/hat.png").returns(io.BytesIO(image_bytes))
+        
+        content_types = context.mock()
+        funk.allows(content_types).find_content_type("word/media/hat.png").returns("image/png")
+        
+        image = _read_and_get_document_xml_element(
+            imagedata_element,
+            content_types=content_types,
+            relationships=relationships,
+            docx_file=docx_file,
+        )
+        assert_equal(documents.Image, type(image))
+        assert_equal("It's a hat", image.alt_text)
+        assert_equal("image/png", image.content_type)
+        with image.open() as image_file:
+            assert_equal(image_bytes, image_file.read())
+        
+        
+    @istest
+    @funk.with_context
     def can_read_inline_pictures(self, context):
         drawing_element = _create_inline_image(
             blip=_embedded_blip("rId5"),