Split off separate HTML writer

1 files changed, 42 insertions, 0 deletions

diff --git a/mammoth/writers/html.py b/mammoth/writers/html.py
new file mode 100644
index 0000000..cf936cd
--- /dev/null
+++ b/mammoth/writers/html.py
@@ -0,0 +1,42 @@
+from __future__ import unicode_literals
+
+import cgi
+
+
+class HtmlWriter(object):
+    def __init__(self):
+        self._fragments = []
+    
+    def text(self, text):
+        self._fragments.append(_escape_html(text))
+    
+    def start(self, name, attributes=None):
+        attribute_string = _generate_attribute_string(attributes)
+        self._fragments.append("<{0}{1}>".format(name, attribute_string))
+
+    def end(self, name):
+        self._fragments.append("</{0}>".format(name))
+    
+    def self_closing(self, name, attributes=None):
+        attribute_string = _generate_attribute_string(attributes)
+        self._fragments.append("<{0}{1} />".format(name, attribute_string))
+    
+    def append(self, html):
+        self._fragments.append(html)
+    
+    def as_string(self):
+        return "".join(self._fragments)
+
+
+def _escape_html(text):
+    return cgi.escape(text, quote=True)
+
+
+def _generate_attribute_string(attributes):
+    if attributes is None:
+        return ""
+    else:
+        return "".join(
+            ' {0}="{1}"'.format(key, _escape_html(attributes[key]))
+            for key in sorted(attributes)
+        )