From 8661cad2a82b66254fff564f4a514cbd67ad371d Mon Sep 17 00:00:00 2001
From: Michael Williamson <mike@zwobble.org>
Date: Sat, 10 Dec 2016 22:01:20 +0000
Subject: Remove self-closing elements from AST

---
 mammoth/conversion.py           |  2 +-
 mammoth/html/__init__.py        | 29 ++++++++---------------------
 mammoth/html/nodes.py           |  7 +++----
 mammoth/images.py               |  2 +-
 tests/conversion_tests.py       |  2 +-
 tests/html/strip_empty_tests.py |  4 ++--
 6 files changed, 16 insertions(+), 30 deletions(-)

diff --git a/mammoth/conversion.py b/mammoth/conversion.py
index bbb4f86..9398c5c 100644
--- a/mammoth/conversion.py
+++ b/mammoth/conversion.py
@@ -179,7 +179,7 @@ class _DocumentConverter(documents.ElementVisitor):
     
     
     def visit_line_break(self, line_break):
-        return [html.self_closing_element("br")]
+        return [html.element("br")]
     
     def visit_note_reference(self, note_reference):
         self._note_references.append(note_reference)
diff --git a/mammoth/html/__init__.py b/mammoth/html/__init__.py
index 6e97ead..b397a81 100644
--- a/mammoth/html/__init__.py
+++ b/mammoth/html/__init__.py
@@ -1,5 +1,5 @@
 from ..lists import flat_map
-from .nodes import TextNode, Tag, Element, SelfClosingElement, ForceWrite, NodeVisitor
+from .nodes import TextNode, Tag, Element, ForceWrite, NodeVisitor
 
 
 def text(value):
@@ -24,13 +24,6 @@ def element(tag_names, attributes=None, children=None, collapsible=None, separat
 
 def collapsible_element(tag_names, attributes=None, children=None):
     return element(tag_names, attributes, children, collapsible=True)
-    
-
-def self_closing_element(tag_name, attributes=None):
-    if attributes is None:
-        attributes = {}
-    
-    return SelfClosingElement(tag_name, attributes)
 
 
 force_write = ForceWrite()
@@ -53,14 +46,11 @@ class StripEmpty(NodeVisitor):
     
     def visit_element(self, element):
         children = strip_empty(element.children)
-        if len(children) == 0:
+        if len(children) == 0 and not element.is_void():
             return []
         else:
             return [Element(element.tag, children)]
     
-    def visit_self_closing_element(self, element):
-        return [element]
-    
     def visit_force_write(self, node):
         return [node]
 
@@ -80,9 +70,6 @@ class _CollapseNode(NodeVisitor):
     def visit_element(self, element):
         return Element(element.tag, collapse(element.children))
     
-    def visit_self_closing_element(self, element):
-        return element
-    
     def visit_force_write(self, node):
         return node
     
@@ -133,12 +120,12 @@ class _NodeWriter(NodeVisitor):
         self._writer.text(node.value)
     
     def visit_element(self, element):
-        self._writer.start(element.tag_name, element.attributes)
-        self.visit_all(element.children)
-        self._writer.end(element.tag_name)
-    
-    def visit_self_closing_element(self, element):
-        self._writer.self_closing(element.tag_name, element.attributes)
+        if element.is_void():
+            self._writer.self_closing(element.tag_name, element.attributes)
+        else:
+            self._writer.start(element.tag_name, element.attributes)
+            self.visit_all(element.children)
+            self._writer.end(element.tag_name)
     
     def visit_force_write(self, element):
         pass
diff --git a/mammoth/html/nodes.py b/mammoth/html/nodes.py
index ff31c8c..9c07259 100644
--- a/mammoth/html/nodes.py
+++ b/mammoth/html/nodes.py
@@ -47,11 +47,10 @@ class Element(Node):
     def separator(self):
         return self.tag.separator
 
+    _VOID_TAG_NAMES = set(["br", "hr", "img"])
 
-@cobble.data
-class SelfClosingElement(Node):
-    tag_name = cobble.field()
-    attributes = cobble.field()
+    def is_void(self):
+        return not self.children and self.tag_name in self._VOID_TAG_NAMES
 
 
 @cobble.visitable
diff --git a/mammoth/images.py b/mammoth/images.py
index 4dc095a..297b0a2 100644
--- a/mammoth/images.py
+++ b/mammoth/images.py
@@ -7,7 +7,7 @@ def img_element(func):
         if image.alt_text:
             attributes["alt"] = image.alt_text
             
-        return [html.self_closing_element("img", attributes)]
+        return [html.element("img", attributes)]
     
     return convert_image
 
diff --git a/tests/conversion_tests.py b/tests/conversion_tests.py
index 0d08b20..71fbbc7 100644
--- a/tests/conversion_tests.py
+++ b/tests/conversion_tests.py
@@ -389,7 +389,7 @@ def images_have_alt_tags_if_available():
 def can_define_custom_conversion_for_images():
     def convert_image(image):
         with image.open() as image_file:
-            return [html.self_closing_element("img", {"alt": image_file.read().decode("ascii")})]
+            return [html.element("img", {"alt": image_file.read().decode("ascii")})]
         
     image = documents.image(alt_text=None, content_type="image/png", open=lambda: io.BytesIO(b"abc"))
     result = convert_document_element_to_html(image, convert_image=convert_image)
diff --git a/tests/html/strip_empty_tests.py b/tests/html/strip_empty_tests.py
index eb28f30..b72b50a 100644
--- a/tests/html/strip_empty_tests.py
+++ b/tests/html/strip_empty_tests.py
@@ -54,8 +54,8 @@ def empty_children_are_removed():
 @istest
 def self_closing_elements_are_never_empty():
     assert_equal(
-        [html.self_closing_element("br")],
-        html.strip_empty([html.self_closing_element("br")]))
+        [html.element("br")],
+        html.strip_empty([html.element("br")]))
 
 
 @istest
-- 
cgit v1.2.3