Write image height and width in HTML

Author: aarbouin

mammoth/images.py

@@ -8,6 +8,9 @@ def convert_image(image):
         attributes = func(image).copy()
         if image.alt_text:
             attributes["alt"] = image.alt_text
+        if image.size:
+            attributes["width"] = image.size.width
+            attributes["height"] = image.size.height
 
         return [html.element("img", attributes)]
 

tests/cli_tests.py

@@ -36,7 +36,7 @@ def html_is_written_to_file_if_output_file_is_set():
 def inline_images_are_included_in_output_if_writing_to_single_file():
     docx_path = test_path("tiny-picture.docx")
     result = _local.run(["mammoth", docx_path])
-    assert_equal(b"""<p><img src="data:image/png;base64,""" + _image_base_64 + b"""" /></p>""", result.output)
+    assert_equal(b"""<p><img height="10" src="data:image/png;base64,""" + _image_base_64 + b"""" width="10" /></p>""", result.output)
 
 
 @istest
@@ -50,7 +50,7 @@ def images_are_written_to_separate_files_if_output_dir_is_set():
         assert_equal(b"", result.stderr_output)
         assert_equal(b"", result.output)
         with open(output_path) as output_file:
-            assert_equal("""<p><img src="1.png" /></p>""", output_file.read())
+            assert_equal("""<p><img height="10" src="1.png" width="10" /></p>""", output_file.read())
 
         with open(image_path, "rb") as image_file:
             assert_equal(_image_base_64, base64.b64encode(image_file.read()))

tests/conversion_tests.py

@@ -510,6 +510,19 @@ def images_have_alt_tags_if_available():
     image_html = parse_xml(io.StringIO(result.value))
     assert_equal('It\'s a hat', image_html.attributes["alt"])
 
+@istest
+def images_have_width_and_height_tags_if_available():
+    image = documents.image(
+        alt_text=None,
+        content_type="image/png",
+        size=documents.Size(width=42, height=51),
+        open=lambda: io.BytesIO(b"abc")
+    )
+    result = convert_document_element_to_html(image)
+    image_html = parse_xml(io.StringIO(result.value))
+    assert_equal('42', image_html.attributes["width"])
+    assert_equal('51', image_html.attributes["height"])
+
 
 @istest
 def can_define_custom_conversion_for_images():

tests/images_tests.py

@@ -17,11 +17,16 @@ def data_uri_encodes_images_in_base64():
     image = mammoth.documents.Image(
         alt_text=None,
         content_type="image/jpeg",
+        size=mammoth.documents.Size(width=800, height=600),
         open=lambda: io.BytesIO(image_bytes),
     )
 
     result = mammoth.images.data_uri(image)
 
     assert_that(result, contains(
-        has_properties(attributes={"src": "data:image/jpeg;base64,YWJj"}),
+        has_properties(attributes={
+            "src": "data:image/jpeg;base64,YWJj",
+            "width": "800",
+            "height": "600",
+        }),
     ))

tests/mammoth_tests.py

@@ -112,23 +112,23 @@ def warning_if_style_mapping_is_not_understood():
 def inline_images_referenced_by_path_relative_to_part_are_included_in_output():
     with open(test_path("tiny-picture.docx"), "rb") as fileobj:
         result = mammoth.convert_to_html(fileobj=fileobj)
-        assert_equal("""<p><img src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAAXNSR0IArs4c6QAAAAlwSFlzAAAOvgAADr4B6kKxwAAAABNJREFUKFNj/M+ADzDhlWUYqdIAQSwBE8U+X40AAAAASUVORK5CYII=" /></p>""", result.value)
+        assert_equal("""<p><img height="10" src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAAXNSR0IArs4c6QAAAAlwSFlzAAAOvgAADr4B6kKxwAAAABNJREFUKFNj/M+ADzDhlWUYqdIAQSwBE8U+X40AAAAASUVORK5CYII=" width="10" /></p>""", result.value)
         assert_equal([], result.messages)
 
 
 @istest
 def inline_images_referenced_by_path_relative_to_base_are_included_in_output():
     with open(test_path("tiny-picture-target-base-relative.docx"), "rb") as fileobj:
         result = mammoth.convert_to_html(fileobj=fileobj)
-        assert_equal("""<p><img src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAAXNSR0IArs4c6QAAAAlwSFlzAAAOvgAADr4B6kKxwAAAABNJREFUKFNj/M+ADzDhlWUYqdIAQSwBE8U+X40AAAAASUVORK5CYII=" /></p>""", result.value)
+        assert_equal("""<p><img height="10" src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAAXNSR0IArs4c6QAAAAlwSFlzAAAOvgAADr4B6kKxwAAAABNJREFUKFNj/M+ADzDhlWUYqdIAQSwBE8U+X40AAAAASUVORK5CYII=" width="10" /></p>""", result.value)
         assert_equal([], result.messages)
 
 
 @istest
 def images_stored_outside_of_document_are_included_in_output():
     with open(test_path("external-picture.docx"), "rb") as fileobj:
         result = mammoth.convert_to_html(fileobj=fileobj)
-        assert_equal("""<p><img src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAAXNSR0IArs4c6QAAAAlwSFlzAAAOvgAADr4B6kKxwAAAABNJREFUKFNj/M+ADzDhlWUYqdIAQSwBE8U+X40AAAAASUVORK5CYII=" /></p>""", result.value)
+        assert_equal("""<p><img height="10" src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAAXNSR0IArs4c6QAAAAlwSFlzAAAOvgAADr4B6kKxwAAAABNJREFUKFNj/M+ADzDhlWUYqdIAQSwBE8U+X40AAAAASUVORK5CYII=" width="10" /></p>""", result.value)
         assert_equal([], result.messages)
 
 
@@ -173,7 +173,7 @@ def convert_image(image):
 
     with open(test_path("tiny-picture.docx"), "rb") as fileobj:
         result = mammoth.convert_to_html(fileobj=fileobj, convert_image=convert_image)
-        assert_equal("""<p><img src="iV,image/png" /></p>""", result.value)
+        assert_equal("""<p><img height="10" src="iV,image/png" width="10" /></p>""", result.value)
         assert_equal([], result.messages)