From 00cfd0d5bae2e7688754dfc458016d42abfcd037 Mon Sep 17 00:00:00 2001
From: Michael Williamson <mike@zwobble.org>
Date: Mon, 14 Mar 2016 08:10:01 +0000
Subject: Add HTML AST with strip_empty function

---
 mammoth/html/__init__.py | 58 ++++++++++++++++++++++++++++++++++++++++++++++++
 mammoth/html/nodes.py    | 32 ++++++++++++++++++++++++++
 2 files changed, 90 insertions(+)
 create mode 100644 mammoth/html/__init__.py
 create mode 100644 mammoth/html/nodes.py

(limited to 'mammoth')

diff --git a/mammoth/html/__init__.py b/mammoth/html/__init__.py
new file mode 100644
index 0000000..905b423
--- /dev/null
+++ b/mammoth/html/__init__.py
@@ -0,0 +1,58 @@
+from ..lists import flat_map
+from .nodes import TextNode, Element, SelfClosingElement, ForceWrite, NodeVisitor
+
+
+def text(value):
+    return TextNode(value)
+
+
+def element(tag_name, attributes=None, children=None):
+    if attributes is None:
+        attributes = {}
+    if children is None:
+        children = []
+    
+    return Element([tag_name], attributes, children, collapsible=False)
+
+
+def self_closing_element(tag_name, attributes=None):
+    if attributes is None:
+        attributes = {}
+    
+    return SelfClosingElement(tag_name, attributes)
+
+
+force_write = ForceWrite()
+
+
+def strip_empty(nodes):
+    return flat_map(_strip_empty_node, nodes)
+
+
+def _strip_empty_node(node):
+    return StripEmpty().visit(node)
+
+
+class StripEmpty(NodeVisitor):
+    def visit_text_node(self, node):
+        if node.value:
+            return [node]
+        else:
+            return []
+    
+    def visit_element(self, element):
+        children = strip_empty(element.children)
+        if len(children) == 0:
+            return []
+        else:
+            return [Element(
+                element.tag_names,
+                element.attributes,
+                children,
+                collapsible=element.collapsible)]
+    
+    def visit_self_closing_element(self, element):
+        return [element]
+    
+    def visit_force_write(self, node):
+        return [node]
diff --git a/mammoth/html/nodes.py b/mammoth/html/nodes.py
new file mode 100644
index 0000000..aae4888
--- /dev/null
+++ b/mammoth/html/nodes.py
@@ -0,0 +1,32 @@
+import cobble
+
+
+class Node(object):
+    pass
+
+
+@cobble.data
+class TextNode(Node):
+    value = cobble.field()
+
+
+@cobble.data
+class Element(Node):
+    tag_names = cobble.field()
+    attributes = cobble.field()
+    children = cobble.field()
+    collapsible = cobble.field()
+
+
+@cobble.data
+class SelfClosingElement(Node):
+    tag_name = cobble.field()
+    attributes = cobble.field()
+
+
+@cobble.visitable
+class ForceWrite(Node):
+    pass
+
+
+NodeVisitor = cobble.visitor(Node)
-- 
cgit v1.2.3