new file mode 100644
--- /dev/null
+++ b/markup/input.py
@@ -0,0 +1,202 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2006 Christopher Lenz
+# All rights reserved.
+#
+# This software is licensed as described in the file COPYING, which
+# you should have received as part of this distribution. The terms
+# are also available at http://trac.edgewall.com/license.html.
+#
+# This software consists of voluntary contributions made by many
+# individuals. For the exact contribution history, see the revision
+# history and logs, available at http://projects.edgewall.com/trac/.
+
+from xml.parsers import expat
+try:
+    frozenset
+except NameError:
+    from sets import ImmutableSet as frozenset
+import HTMLParser as html
+import htmlentitydefs
+import re
+from StringIO import StringIO
+
+from markup.core import Attributes, Markup, QName, Stream
+
+
+class XMLParser(object):
+    """Generator-based XML parser based on roughly equivalent code in
+    Kid/ElementTree."""
+
+    def __init__(self, source):
+        self.source = source
+
+        # Setup the Expat parser
+        parser = expat.ParserCreate('utf-8', '}')
+        parser.buffer_text = True
+        parser.returns_unicode = True
+        parser.StartElementHandler = self._handle_start
+        parser.EndElementHandler = self._handle_end
+        parser.CharacterDataHandler = self._handle_data
+        parser.XmlDeclHandler = self._handle_prolog
+        parser.StartDoctypeDeclHandler = self._handle_doctype
+        parser.StartNamespaceDeclHandler = self._handle_start_ns
+        parser.EndNamespaceDeclHandler = self._handle_end_ns
+        parser.ProcessingInstructionHandler = self._handle_pi
+        parser.CommentHandler = self._handle_comment
+        parser.DefaultHandler = self._handle_other
+
+        # Location reporting is only support in Python >= 2.4
+        if not hasattr(parser, 'CurrentLineNumber'):
+            self.getpos = self._getpos_unknown
+
+        self.expat = parser
+        self.queue = []
+
+    def __iter__(self):
+        bufsize = 4 * 1024 # 4K
+        done = False
+        while True:
+            while not done and len(self.queue) == 0:
+                data = self.source.read(bufsize)
+                if data == '': # end of data
+                    if hasattr(self, 'expat'):
+                        self.expat.Parse('', True)
+                        del self.expat # get rid of circular references
+                    done = True
+                else:
+                    self.expat.Parse(data, False)
+            for event in self.queue:
+                yield event
+            self.queue = []
+            if done:
+                break
+
+    def _getpos_unknown(self):
+        return (-1, -1)
+
+    def getpos(self):
+        return self.expat.CurrentLineNumber, self.expat.CurrentColumnNumber
+
+    def _handle_start(self, tag, attrib):
+        self.queue.append((Stream.START, (QName(tag), Attributes(attrib.items())),
+                           self.getpos()))
+
+    def _handle_end(self, tag):
+        self.queue.append((Stream.END, QName(tag), self.getpos()))
+
+    def _handle_data(self, text):
+        self.queue.append((Stream.TEXT, text, self.getpos()))
+
+    def _handle_prolog(self, version, encoding, standalone):
+        self.queue.append((Stream.PROLOG, (version, encoding, standalone),
+                           self.getpos()))
+
+    def _handle_doctype(self, name, sysid, pubid, has_internal_subset):
+        self.queue.append((Stream.DOCTYPE, (name, pubid, sysid), self.getpos()))
+
+    def _handle_start_ns(self, prefix, uri):
+        self.queue.append((Stream.START_NS, (prefix or '', uri), self.getpos()))
+
+    def _handle_end_ns(self, prefix):
+        self.queue.append((Stream.END_NS, prefix or '', self.getpos()))
+
+    def _handle_pi(self, target, data):
+        self.queue.append((Stream.PI, (target, data), self.getpos()))
+
+    def _handle_comment(self, text):
+        self.queue.append((Stream.COMMENT, text, self.getpos()))
+
+    def _handle_other(self, text):
+        if text.startswith('&'):
+            # deal with undefined entities
+            try:
+                text = unichr(htmlentitydefs.name2codepoint[text[1:-1]])
+                self.queue.append((Stream.TEXT, text, self.getpos()))
+            except KeyError:
+                lineno, offset = self.getpos()
+                raise expat.error("undefined entity %s: line %d, column %d" %
+                                  (text, lineno, offset))
+
+
+def XML(text):
+    return Stream(list(XMLParser(StringIO(text))))
+
+
+class HTMLParser(html.HTMLParser):
+    """Parser for HTML input based on the Python `HTMLParser` module.
+    
+    This class provides the same interface for generating stream events as
+    `XMLParser`, and attempts to automatically balance tags.
+    """
+
+    _EMPTY_ELEMS = frozenset(['area', 'base', 'basefont', 'br', 'col', 'frame',
+                              'hr', 'img', 'input', 'isindex', 'link', 'meta',
+                              'param'])
+
+    def __init__(self, source):
+        html.HTMLParser.__init__(self)
+        self.source = source
+        self.queue = []
+        self._open_tags = []
+
+    def __iter__(self):
+        bufsize = 4 * 1024 # 4K
+        done = False
+        while True:
+            while not done and len(self.queue) == 0:
+                data = self.source.read(bufsize)
+                if data == '': # end of data
+                    self.close()
+                    done = True
+                else:
+                    self.feed(data)
+            for kind, data, pos in self.queue:
+                yield kind, data, pos
+            self.queue = []
+            if done:
+                open_tags = self._open_tags
+                open_tags.reverse()
+                for tag in open_tags:
+                    yield Stream.END, QName(tag), pos
+                break
+
+    def handle_starttag(self, tag, attrib):
+        pos = self.getpos()
+        self.queue.append((Stream.START, (QName(tag), Attributes(attrib)), pos))
+        if tag in self._EMPTY_ELEMS:
+            self.queue.append((Stream.END, QName(tag), pos))
+        else:
+            self._open_tags.append(tag)
+
+    def handle_endtag(self, tag):
+        if tag not in self._EMPTY_ELEMS:
+            pos = self.getpos()
+            while self._open_tags:
+                open_tag = self._open_tags.pop()
+                if open_tag.lower() == tag.lower():
+                    break
+                self.queue.append((Stream.END, QName(open_tag), pos))
+            self.queue.append((Stream.END, QName(tag), pos))
+
+    def handle_data(self, text):
+        self.queue.append((Stream.TEXT, text, self.getpos()))
+
+    def handle_charref(self, name):
+        self.queue.append((Stream.TEXT, Markup('&#%s;' % name), self.getpos()))
+
+    def handle_entityref(self, name):
+        self.queue.append((Stream.TEXT, Markup('&%s;' % name), self.getpos()))
+
+    def handle_pi(self, data):
+        target, data = data.split(maxsplit=1)
+        data = data.rstrip('?')
+        self.queue.append((Stream.PI, (target.strip(), data.strip()),
+                           self.getpos()))
+
+    def handle_comment(self, text):
+        self.queue.append((Stream.COMMENT, text, self.getpos()))
+
+
+def HTML(text):
+    return Stream(list(HTMLParser(StringIO(text))))