Add HTML AST with strip_empty function

4 files changed, 156 insertions, 0 deletions

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)
diff --git a/tests/html/__init__.py b/tests/html/__init__.py
new file mode 100644
index 0000000..8b13789
--- /dev/null
+++ b/tests/html/__init__.py
@@ -0,0 +1 @@
+
diff --git a/tests/html/strip_empty_tests.py b/tests/html/strip_empty_tests.py
new file mode 100644
index 0000000..eb28f30
--- /dev/null
+++ b/tests/html/strip_empty_tests.py
@@ -0,0 +1,65 @@
+from nose.tools import istest, assert_equal
+
+from mammoth import html
+
+
+@istest
+def text_nodes_with_text_are_not_stripped():
+    assert_equal(
+        [html.text("H")],
+        html.strip_empty([html.text("H")]))
+
+
+@istest
+def empty_text_nodes_are_stripped():
+    assert_equal(
+        [],
+        html.strip_empty([html.text("")]))
+
+
+@istest
+def elements_with_non_empty_children_are_not_stripped():
+    assert_equal(
+        [html.element("p", {}, [html.text("H")])],
+        html.strip_empty([html.element("p", {}, [html.text("H")])]))
+
+
+@istest
+def elements_with_no_children_are_stripped():
+    assert_equal(
+        [],
+        html.strip_empty([html.element("p")]))
+
+
+@istest
+def elements_with_only_empty_children_are_stripped():
+    assert_equal(
+        [],
+        html.strip_empty([html.element("p", {}, [html.text("")])]))
+
+
+@istest
+def empty_children_are_removed():
+    assert_equal(
+        html.strip_empty([html.element("ul", {}, [
+            html.element("li", {}, [html.text("")]),
+            html.element("li", {}, [html.text("H")]),
+        ])]),
+        
+        [html.element("ul", {}, [
+            html.element("li", {}, [html.text("H")])
+        ])])
+
+
+@istest
+def self_closing_elements_are_never_empty():
+    assert_equal(
+        [html.self_closing_element("br")],
+        html.strip_empty([html.self_closing_element("br")]))
+
+
+@istest
+def force_writes_are_never_empty():
+    assert_equal(
+        [html.force_write],
+        html.strip_empty([html.force_write]))