Update html5lib 1.1-dev (05b73ef) → 1.1 (f87487a).

CHANGES.md

@@ -23,7 +23,7 @@
 * Update feedparser 6.0.0b1 (d12d3bd) to feedparser_py2 6.0.0b3 (7e255f0)
 * Add feedparser_py3 6.0.0b3 (7e255f0)
 * Update Fuzzywuzzy 0.17.0 (0cfb2c8) to 0.18.0 (2188520)
-* Update html5lib 1.1-dev (4b22754) to 1.1-dev (05b73ef)
+* Update html5lib 1.1-dev (4b22754) to 1.1 (f87487a)
 * Update idna library 2.8 (032fc55) to 2.9 (1233a73)
 * Update isotope library 3.0.1 (98ba374) to 3.0.6 (ad00807)
 * Update functools_lru_cache 1.5 (21e85f5) to 1.6.1 (2dc65b5)
@@ -62,6 +62,8 @@
 * Update Beautiful Soup 4.8.2 (r544) to 4.8.2 (r556)
 * Update Certifi 2019.06.16 (84dc766) to 2019.11.28 (21abb9b)
 * Update dateutil 2.8.1 (fc9b162) to 2.8.1 (110a09b)
+* Update html5lib 1.1-dev (4b22754) to 1.1-dev (05b73ef)
+* Update html5lib 1.1-dev (05b73ef) to 1.1 (f87487a)
 * Update Requests library 2.22.0 (3d968ff) to 2.22.0 (d2f65af)
 * Update Requests library 2.22.0 (d2f65af) to 2.23.0 (b7c6aba)
 * Update Requests library 2.23.0 (b7c6aba) to 2.24.0 (1b41763)

lib/html5lib/__init__.py

@@ -32,4 +32,4 @@ __all__ = ["HTMLParser", "parse", "parseFragment", "getTreeBuilder",
 
 # this has to be at the top level, see how setup.py parses this
 #: Distribution version number.
-__version__ = "1.1-dev"
+__version__ = "1.1"

lib/html5lib/_inputstream.py

@@ -658,9 +658,7 @@ class EncodingBytes(bytes):
         """Look for a sequence of bytes at the start of a string. If the bytes
         are found return True and advance the position to the byte after the
         match. Otherwise return False and leave the position alone"""
-        p = self.position
-        data = self[p:p + len(bytes)]
-        rv = data.startswith(bytes)
+        rv = self.startswith(bytes, self.position)
         if rv:
             self.position += len(bytes)
         return rv
@@ -668,15 +666,11 @@ class EncodingBytes(bytes):
     def jumpTo(self, bytes):
         """Look for the next sequence of bytes matching a given sequence. If
         a match is found advance the position to the last byte of the match"""
-        newPosition = self[self.position:].find(bytes)
-        if newPosition > -1:
-            # XXX: This is ugly, but I can't see a nicer way to fix this.
-            if self._position == -1:
-                self._position = 0
-            self._position += (newPosition + len(bytes) - 1)
-            return True
-        else:
+        try:
+            self._position = self.index(bytes, self.position) + len(bytes) - 1
+        except ValueError:
             raise StopIteration
+        return True
 
 
 class EncodingParser(object):
@@ -688,6 +682,9 @@ class EncodingParser(object):
         self.encoding = None
 
     def getEncoding(self):
+        if b"<meta" not in self.data:
+            return None
+
         methodDispatch = (
             (b"<!--", self.handleComment),
             (b"<meta", self.handleMeta),
@@ -697,6 +694,10 @@ class EncodingParser(object):
             (b"<", self.handlePossibleStartTag))
         for _ in self.data:
             keepParsing = True
+            try:
+                self.data.jumpTo(b"<")
+            except StopIteration:
+                break
             for key, method in methodDispatch:
                 if self.data.matchBytes(key):
                     try:

lib/html5lib/_tokenizer.py

@@ -2,7 +2,8 @@ from __future__ import absolute_import, division, unicode_literals
 
 from six import unichr as chr
 
-from collections import deque
+from collections import deque, OrderedDict
+from sys import version_info
 
 from .constants import spaceCharacters
 from .constants import entities
@@ -17,6 +18,11 @@ from ._trie import Trie
 
 entitiesTrie = Trie(entities)
 
+if version_info >= (3, 7):
+    attributeMap = dict
+else:
+    attributeMap = OrderedDict
+
 
 class HTMLTokenizer(object):
     """ This class takes care of tokenizing HTML.
@@ -228,6 +234,14 @@ class HTMLTokenizer(object):
         # Add token to the queue to be yielded
         if (token["type"] in tagTokenTypes):
             token["name"] = token["name"].translate(asciiUpper2Lower)
+            if token["type"] == tokenTypes["StartTag"]:
+                raw = token["data"]
+                data = attributeMap(raw)
+                if len(raw) > len(data):
+                    # we had some duplicated attribute, fix so first wins
+                    data.update(raw[::-1])
+                token["data"] = data
+
             if token["type"] == tokenTypes["EndTag"]:
                 if token["data"]:
                     self.tokenQueue.append({"type": tokenTypes["ParseError"],

lib/html5lib/_trie/__init__.py

@@ -1,14 +1,5 @@
 from __future__ import absolute_import, division, unicode_literals
 
-from .py import Trie as PyTrie
+from .py import Trie
 
-Trie = PyTrie
-
-# pylint:disable=wrong-import-position
-try:
-    from .datrie import Trie as DATrie
-except ImportError:
-    pass
-else:
-    Trie = DATrie
-# pylint:enable=wrong-import-position
+__all__ = ["Trie"]

lib/html5lib/_trie/datrie.py

@@ -1,44 +0,0 @@
-from __future__ import absolute_import, division, unicode_literals
-
-from datrie import Trie as DATrie
-from six import text_type
-
-from ._base import Trie as ABCTrie
-
-
-class Trie(ABCTrie):
-    def __init__(self, data):
-        chars = set()
-        for key in data.keys():
-            if not isinstance(key, text_type):
-                raise TypeError("All keys must be strings")
-            for char in key:
-                chars.add(char)
-
-        self._data = DATrie("".join(chars))
-        for key, value in data.items():
-            self._data[key] = value
-
-    def __contains__(self, key):
-        return key in self._data
-
-    def __len__(self):
-        return len(self._data)
-
-    def __iter__(self):
-        raise NotImplementedError()
-
-    def __getitem__(self, key):
-        return self._data[key]
-
-    def keys(self, prefix=None):
-        return self._data.keys(prefix)
-
-    def has_keys_with_prefix(self, prefix):
-        return self._data.has_keys_with_prefix(prefix)
-
-    def longest_prefix(self, prefix):
-        return self._data.longest_prefix(prefix)
-
-    def longest_prefix_item(self, prefix):
-        return self._data.longest_prefix_item(prefix)

lib/html5lib/_utils.py

@@ -2,11 +2,19 @@ from __future__ import absolute_import, division, unicode_literals
 
 from types import ModuleType
 
-from six import text_type
-
 try:
-    import xml.etree.cElementTree as default_etree
+    from collections.abc import Mapping
 except ImportError:
+    from collections import Mapping
+
+from six import text_type, PY3
+
+if PY3:
+    import xml.etree.ElementTree as default_etree
+else:
+    try:
+        import xml.etree.cElementTree as default_etree
+    except ImportError:
         import xml.etree.ElementTree as default_etree
 
 
@@ -47,9 +55,6 @@ class MethodDispatcher(dict):
     """
 
     def __init__(self, items=()):
-        # Using _dictEntries instead of directly assigning to self is about
-        # twice as fast. Please do careful performance testing before changing
-        # anything here.
         _dictEntries = []
         for name, value in items:
             if isinstance(name, (list, tuple, frozenset, set)):
@@ -64,6 +69,36 @@ class MethodDispatcher(dict):
     def __getitem__(self, key):
         return dict.get(self, key, self.default)
 
+    def __get__(self, instance, owner=None):
+        return BoundMethodDispatcher(instance, self)
+
+
+class BoundMethodDispatcher(Mapping):
+    """Wraps a MethodDispatcher, binding its return values to `instance`"""
+    def __init__(self, instance, dispatcher):
+        self.instance = instance
+        self.dispatcher = dispatcher
+
+    def __getitem__(self, key):
+        # see https://docs.python.org/3/reference/datamodel.html#object.__get__
+        # on a function, __get__ is used to bind a function to an instance as a bound method
+        return self.dispatcher[key].__get__(self.instance)
+
+    def get(self, key, default):
+        if key in self.dispatcher:
+            return self[key]
+        else:
+            return default
+
+    def __iter__(self):
+        return iter(self.dispatcher)
+
+    def __len__(self):
+        return len(self.dispatcher)
+
+    def __contains__(self, key):
+        return key in self.dispatcher
+
 
 # Some utility functions to deal with weirdness around UCS2 vs UCS4
 # python builds

lib/html5lib/filters/sanitizer.py

@@ -1,6 +1,15 @@
+"""Deprecated from html5lib 1.1.
+
+See `here <https://github.com/html5lib/html5lib-python/issues/443>`_ for
+information about its deprecation; `Bleach <https://github.com/mozilla/bleach>`_
+is recommended as a replacement. Please let us know in the aforementioned issue
+if Bleach is unsuitable for your needs.
+
+"""
 from __future__ import absolute_import, division, unicode_literals
 
 import re
+import warnings
 from xml.sax.saxutils import escape, unescape
 
 from six.moves import urllib_parse as urlparse
@@ -11,6 +20,14 @@ from ..constants import namespaces, prefixes
 __all__ = ["Filter"]
 
 
+_deprecation_msg = (
+    "html5lib's sanitizer is deprecated; see " +
+    "https://github.com/html5lib/html5lib-python/issues/443 and please let " +
+    "us know if Bleach is unsuitable for your needs"
+)
+
+warnings.warn(_deprecation_msg, DeprecationWarning)
+
 allowed_elements = frozenset((
     (namespaces['html'], 'a'),
     (namespaces['html'], 'abbr'),
@@ -750,6 +767,9 @@ class Filter(base.Filter):
 
         """
         super(Filter, self).__init__(source)
+
+        warnings.warn(_deprecation_msg, DeprecationWarning)
+
         self.allowed_elements = allowed_elements
         self.allowed_attributes = allowed_attributes
         self.allowed_css_properties = allowed_css_properties

lib/html5lib/html5parser.py

@@ -2,7 +2,6 @@ from __future__ import absolute_import, division, unicode_literals
 from six import with_metaclass, viewkeys
 
 import types
-from collections import OrderedDict
 
 from . import _inputstream
 from . import _tokenizer
@@ -202,7 +201,7 @@ class HTMLParser(object):
         DoctypeToken = tokenTypes["Doctype"]
         ParseErrorToken = tokenTypes["ParseError"]
 
-        for token in self.normalizedTokens():
+        for token in self.tokenizer:
             prev_token = None
             new_token = token
             while new_token is not None:
@@ -260,10 +259,6 @@ class HTMLParser(object):
             if reprocess:
                 assert self.phase not in phases
 
-    def normalizedTokens(self):
-        for token in self.tokenizer:
-            yield self.normalizeToken(token)
-
     def parse(self, stream, *args, **kwargs):
         """Parse a HTML document into a well-formed tree
 
@@ -325,17 +320,6 @@ class HTMLParser(object):
         if self.strict:
             raise ParseError(E[errorcode] % datavars)
 
-    def normalizeToken(self, token):
-        # HTML5 specific normalizations to the token stream
-        if token["type"] == tokenTypes["StartTag"]:
-            raw = token["data"]
-            token["data"] = OrderedDict(raw)
-            if len(raw) > len(token["data"]):
-                # we had some duplicated attribute, fix so first wins
-                token["data"].update(raw[::-1])
-
-        return token
-
     def adjustMathMLAttributes(self, token):
         adjust_attributes(token, adjustMathMLAttributes)
 
@@ -442,10 +426,13 @@ def getPhases(debug):
     class Phase(with_metaclass(getMetaclass(debug, log))):
         """Base class for helper object that implements each phase of processing
         """
+        __slots__ = ("parser", "tree", "__startTagCache", "__endTagCache")
 
         def __init__(self, parser, tree):
             self.parser = parser
             self.tree = tree
+            self.__startTagCache = {}
+            self.__endTagCache = {}
 
         def processEOF(self):
             raise NotImplementedError
@@ -465,7 +452,21 @@ def getPhases(debug):
             self.tree.insertText(token["data"])
 
         def processStartTag(self, token):
-            return self.startTagHandler[token["name"]](token)
+            # Note the caching is done here rather than BoundMethodDispatcher as doing it there
+            # requires a circular reference to the Phase, and this ends up with a significant
+            # (CPython 2.7, 3.8) GC cost when parsing many short inputs
+            name = token["name"]
+            # In Py2, using `in` is quicker in general than try/except KeyError
+            # In Py3, `in` is quicker when there are few cache hits (typically short inputs)
+            if name in self.__startTagCache:
+                func = self.__startTagCache[name]
+            else:
+                func = self.__startTagCache[name] = self.startTagHandler[name]
+                # bound the cache size in case we get loads of unknown tags
+                while len(self.__startTagCache) > len(self.startTagHandler) * 1.1:
+                    # this makes the eviction policy random on Py < 3.7 and FIFO >= 3.7
+                    self.__startTagCache.pop(next(iter(self.__startTagCache)))
+            return func(token)
 
         def startTagHtml(self, token):
             if not self.parser.firstStartTag and token["name"] == "html":
@@ -478,9 +479,25 @@ def getPhases(debug):
             self.parser.firstStartTag = False
 
         def processEndTag(self, token):
-            return self.endTagHandler[token["name"]](token)
+            # Note the caching is done here rather than BoundMethodDispatcher as doing it there
+            # requires a circular reference to the Phase, and this ends up with a significant
+            # (CPython 2.7, 3.8) GC cost when parsing many short inputs
+            name = token["name"]
+            # In Py2, using `in` is quicker in general than try/except KeyError
+            # In Py3, `in` is quicker when there are few cache hits (typically short inputs)
+            if name in self.__endTagCache:
+                func = self.__endTagCache[name]
+            else:
+                func = self.__endTagCache[name] = self.endTagHandler[name]
+                # bound the cache size in case we get loads of unknown tags
+                while len(self.__endTagCache) > len(self.endTagHandler) * 1.1:
+                    # this makes the eviction policy random on Py < 3.7 and FIFO >= 3.7
+                    self.__endTagCache.pop(next(iter(self.__endTagCache)))
+            return func(token)
 
     class InitialPhase(Phase):
+        __slots__ = tuple()
+
         def processSpaceCharacters(self, token):
             pass
 
@@ -609,6 +626,8 @@ def getPhases(debug):
             return True
 
     class BeforeHtmlPhase(Phase):
+        __slots__ = tuple()
+
         # helper methods
         def insertHtmlElement(self):
             self.tree.insertRoot(impliedTagToken("html", "StartTag"))
@@ -644,19 +663,7 @@ def getPhases(debug):
                 return token
 
     class BeforeHeadPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("head", self.startTagHead)
-            ])
-            self.startTagHandler.default = self.startTagOther
-
-            self.endTagHandler = _utils.MethodDispatcher([
-                (("head", "body", "html", "br"), self.endTagImplyHead)
-            ])
-            self.endTagHandler.default = self.endTagOther
+        __slots__ = tuple()
 
         def processEOF(self):
             self.startTagHead(impliedTagToken("head", "StartTag"))
@@ -689,28 +696,19 @@ def getPhases(debug):
             self.parser.parseError("end-tag-after-implied-root",
                                    {"name": token["name"]})
 
-    class InHeadPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("title", self.startTagTitle),
-                (("noframes", "style"), self.startTagNoFramesStyle),
-                ("noscript", self.startTagNoscript),
-                ("script", self.startTagScript),
-                (("base", "basefont", "bgsound", "command", "link"),
-                 self.startTagBaseLinkCommand),
-                ("meta", self.startTagMeta),
-                ("head", self.startTagHead)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", startTagHtml),
+            ("head", startTagHead)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("head", self.endTagHead),
-                (("br", "html", "body"), self.endTagHtmlBodyBr)
+        endTagHandler = _utils.MethodDispatcher([
+            (("head", "body", "html", "br"), endTagImplyHead)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InHeadPhase(Phase):
+        __slots__ = tuple()
 
         # the real thing
         def processEOF(self):
@@ -792,22 +790,27 @@ def getPhases(debug):
         def anythingElse(self):
             self.endTagHead(impliedTagToken("head"))
 
-    class InHeadNoscriptPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                (("basefont", "bgsound", "link", "meta", "noframes", "style"), self.startTagBaseLinkCommand),
-                (("head", "noscript"), self.startTagHeadNoscript),
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", startTagHtml),
+            ("title", startTagTitle),
+            (("noframes", "style"), startTagNoFramesStyle),
+            ("noscript", startTagNoscript),
+            ("script", startTagScript),
+            (("base", "basefont", "bgsound", "command", "link"),
+             startTagBaseLinkCommand),
+            ("meta", startTagMeta),
+            ("head", startTagHead)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("noscript", self.endTagNoscript),
-                ("br", self.endTagBr),
+        endTagHandler = _utils.MethodDispatcher([
+            ("head", endTagHead),
+            (("br", "html", "body"), endTagHtmlBodyBr)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InHeadNoscriptPhase(Phase):
+        __slots__ = tuple()
 
         def processEOF(self):
             self.parser.parseError("eof-in-head-noscript")
@@ -856,23 +859,21 @@ def getPhases(debug):
             # Caller must raise parse error first!
             self.endTagNoscript(impliedTagToken("noscript"))
 
-    class AfterHeadPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", startTagHtml),
+            (("basefont", "bgsound", "link", "meta", "noframes", "style"), startTagBaseLinkCommand),
+            (("head", "noscript"), startTagHeadNoscript),
+        ])
+        startTagHandler.default = startTagOther
 
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("body", self.startTagBody),
-                ("frameset", self.startTagFrameset),
-                (("base", "basefont", "bgsound", "link", "meta", "noframes", "script",
-                  "style", "title"),
-                 self.startTagFromHead),
-                ("head", self.startTagHead)
+        endTagHandler = _utils.MethodDispatcher([
+            ("noscript", endTagNoscript),
+            ("br", endTagBr),
         ])
-            self.startTagHandler.default = self.startTagOther
-            self.endTagHandler = _utils.MethodDispatcher([(("body", "html", "br"),
-                                                           self.endTagHtmlBodyBr)])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class AfterHeadPhase(Phase):
+        __slots__ = tuple()
 
         def processEOF(self):
             self.anythingElse()
@@ -923,80 +924,30 @@ def getPhases(debug):
             self.parser.phase = self.parser.phases["inBody"]
             self.parser.framesetOK = True
 
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", startTagHtml),
+            ("body", startTagBody),
+            ("frameset", startTagFrameset),
+            (("base", "basefont", "bgsound", "link", "meta", "noframes", "script",
+              "style", "title"),
+             startTagFromHead),
+            ("head", startTagHead)
+        ])
+        startTagHandler.default = startTagOther
+        endTagHandler = _utils.MethodDispatcher([(("body", "html", "br"),
+                                                  endTagHtmlBodyBr)])
+        endTagHandler.default = endTagOther
+
     class InBodyPhase(Phase):
         # http://www.whatwg.org/specs/web-apps/current-work/#parsing-main-inbody
         # the really-really-really-very crazy mode
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
+        __slots__ = ("processSpaceCharacters",)
 
+        def __init__(self, *args, **kwargs):
+            super(InBodyPhase, self).__init__(*args, **kwargs)
             # Set this to the default handler
             self.processSpaceCharacters = self.processSpaceCharactersNonPre
 
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                (("base", "basefont", "bgsound", "command", "link", "meta",
-                  "script", "style", "title"),
-                 self.startTagProcessInHead),
-                ("body", self.startTagBody),
-                ("frameset", self.startTagFrameset),
-                (("address", "article", "aside", "blockquote", "center", "details",
-                  "dir", "div", "dl", "fieldset", "figcaption", "figure",
-                  "footer", "header", "hgroup", "main", "menu", "nav", "ol", "p",
-                  "section", "summary", "ul"),
-                 self.startTagCloseP),
-                (headingElements, self.startTagHeading),
-                (("pre", "listing"), self.startTagPreListing),
-                ("form", self.startTagForm),
-                (("li", "dd", "dt"), self.startTagListItem),
-                ("plaintext", self.startTagPlaintext),
-                ("a", self.startTagA),
-                (("b", "big", "code", "em", "font", "i", "s", "small", "strike",
-                  "strong", "tt", "u"), self.startTagFormatting),
-                ("nobr", self.startTagNobr),
-                ("button", self.startTagButton),
-                (("applet", "marquee", "object"), self.startTagAppletMarqueeObject),
-                ("xmp", self.startTagXmp),
-                ("table", self.startTagTable),
-                (("area", "br", "embed", "img", "keygen", "wbr"),
-                 self.startTagVoidFormatting),
-                (("param", "source", "track"), self.startTagParamSource),
-                ("input", self.startTagInput),
-                ("hr", self.startTagHr),
-                ("image", self.startTagImage),
-                ("isindex", self.startTagIsIndex),
-                ("textarea", self.startTagTextarea),
-                ("iframe", self.startTagIFrame),
-                ("noscript", self.startTagNoscript),
-                (("noembed", "noframes"), self.startTagRawtext),
-                ("select", self.startTagSelect),
-                (("rp", "rt"), self.startTagRpRt),
-                (("option", "optgroup"), self.startTagOpt),
-                (("math"), self.startTagMath),
-                (("svg"), self.startTagSvg),
-                (("caption", "col", "colgroup", "frame", "head",
-                  "tbody", "td", "tfoot", "th", "thead",
-                  "tr"), self.startTagMisplaced)
-            ])
-            self.startTagHandler.default = self.startTagOther
-
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("body", self.endTagBody),
-                ("html", self.endTagHtml),
-                (("address", "article", "aside", "blockquote", "button", "center",
-                  "details", "dialog", "dir", "div", "dl", "fieldset", "figcaption", "figure",
-                  "footer", "header", "hgroup", "listing", "main", "menu", "nav", "ol", "pre",
-                  "section", "summary", "ul"), self.endTagBlock),
-                ("form", self.endTagForm),
-                ("p", self.endTagP),
-                (("dd", "dt", "li"), self.endTagListItem),
-                (headingElements, self.endTagHeading),
-                (("a", "b", "big", "code", "em", "font", "i", "nobr", "s", "small",
-                  "strike", "strong", "tt", "u"), self.endTagFormatting),
-                (("applet", "marquee", "object"), self.endTagAppletMarqueeObject),
-                ("br", self.endTagBr),
-            ])
-            self.endTagHandler.default = self.endTagOther
-
         def isMatchingFormattingElement(self, node1, node2):
             return (node1.name == node2.name and
                     node1.namespace == node2.namespace and
@@ -1646,14 +1597,73 @@ def getPhases(debug):
                         self.parser.parseError("unexpected-end-tag", {"name": token["name"]})
                         break
 
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            (("base", "basefont", "bgsound", "command", "link", "meta",
+              "script", "style", "title"),
+             startTagProcessInHead),
+            ("body", startTagBody),
+            ("frameset", startTagFrameset),
+            (("address", "article", "aside", "blockquote", "center", "details",
+              "dir", "div", "dl", "fieldset", "figcaption", "figure",
+              "footer", "header", "hgroup", "main", "menu", "nav", "ol", "p",
+              "section", "summary", "ul"),
+             startTagCloseP),
+            (headingElements, startTagHeading),
+            (("pre", "listing"), startTagPreListing),
+            ("form", startTagForm),
+            (("li", "dd", "dt"), startTagListItem),
+            ("plaintext", startTagPlaintext),
+            ("a", startTagA),
+            (("b", "big", "code", "em", "font", "i", "s", "small", "strike",
+              "strong", "tt", "u"), startTagFormatting),
+            ("nobr", startTagNobr),
+            ("button", startTagButton),
+            (("applet", "marquee", "object"), startTagAppletMarqueeObject),
+            ("xmp", startTagXmp),
+            ("table", startTagTable),
+            (("area", "br", "embed", "img", "keygen", "wbr"),
+             startTagVoidFormatting),
+            (("param", "source", "track"), startTagParamSource),
+            ("input", startTagInput),
+            ("hr", startTagHr),
+            ("image", startTagImage),
+            ("isindex", startTagIsIndex),
+            ("textarea", startTagTextarea),
+            ("iframe", startTagIFrame),
+            ("noscript", startTagNoscript),
+            (("noembed", "noframes"), startTagRawtext),
+            ("select", startTagSelect),
+            (("rp", "rt"), startTagRpRt),
+            (("option", "optgroup"), startTagOpt),
+            (("math"), startTagMath),
+            (("svg"), startTagSvg),
+            (("caption", "col", "colgroup", "frame", "head",
+              "tbody", "td", "tfoot", "th", "thead",
+              "tr"), startTagMisplaced)
+        ])
+        startTagHandler.default = startTagOther
+
+        endTagHandler = _utils.MethodDispatcher([
+            ("body", endTagBody),
+            ("html", endTagHtml),
+            (("address", "article", "aside", "blockquote", "button", "center",
+              "details", "dialog", "dir", "div", "dl", "fieldset", "figcaption", "figure",
+              "footer", "header", "hgroup", "listing", "main", "menu", "nav", "ol", "pre",
+              "section", "summary", "ul"), endTagBlock),
+            ("form", endTagForm),
+            ("p", endTagP),
+            (("dd", "dt", "li"), endTagListItem),
+            (headingElements, endTagHeading),
+            (("a", "b", "big", "code", "em", "font", "i", "nobr", "s", "small",
+              "strike", "strong", "tt", "u"), endTagFormatting),
+            (("applet", "marquee", "object"), endTagAppletMarqueeObject),
+            ("br", endTagBr),
+        ])
+        endTagHandler.default = endTagOther
+
     class TextPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-            self.startTagHandler = _utils.MethodDispatcher([])
-            self.startTagHandler.default = self.startTagOther
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("script", self.endTagScript)])
-            self.endTagHandler.default = self.endTagOther
+        __slots__ = tuple()
 
         def processCharacters(self, token):
             self.tree.insertText(token["data"])
@@ -1679,30 +1689,15 @@ def getPhases(debug):
             self.tree.openElements.pop()
             self.parser.phase = self.parser.originalPhase
 
+        startTagHandler = _utils.MethodDispatcher([])
+        startTagHandler.default = startTagOther
+        endTagHandler = _utils.MethodDispatcher([
+            ("script", endTagScript)])
+        endTagHandler.default = endTagOther
+
     class InTablePhase(Phase):
         # http://www.whatwg.org/specs/web-apps/current-work/#in-table
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("caption", self.startTagCaption),
-                ("colgroup", self.startTagColgroup),
-                ("col", self.startTagCol),
-                (("tbody", "tfoot", "thead"), self.startTagRowGroup),
-                (("td", "th", "tr"), self.startTagImplyTbody),
-                ("table", self.startTagTable),
-                (("style", "script"), self.startTagStyleScript),
-                ("input", self.startTagInput),
-                ("form", self.startTagForm)
-            ])
-            self.startTagHandler.default = self.startTagOther
-
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("table", self.endTagTable),
-                (("body", "caption", "col", "colgroup", "html", "tbody", "td",
-                  "tfoot", "th", "thead", "tr"), self.endTagIgnore)
-            ])
-            self.endTagHandler.default = self.endTagOther
+        __slots__ = tuple()
 
         # helper methods
         def clearStackToTableContext(self):
@@ -1824,9 +1819,32 @@ def getPhases(debug):
             self.parser.phases["inBody"].processEndTag(token)
             self.tree.insertFromTable = False
 
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            ("caption", startTagCaption),
+            ("colgroup", startTagColgroup),
+            ("col", startTagCol),
+            (("tbody", "tfoot", "thead"), startTagRowGroup),
+            (("td", "th", "tr"), startTagImplyTbody),
+            ("table", startTagTable),
+            (("style", "script"), startTagStyleScript),
+            ("input", startTagInput),
+            ("form", startTagForm)
+        ])
+        startTagHandler.default = startTagOther
+
+        endTagHandler = _utils.MethodDispatcher([
+            ("table", endTagTable),
+            (("body", "caption", "col", "colgroup", "html", "tbody", "td",
+              "tfoot", "th", "thead", "tr"), endTagIgnore)
+        ])
+        endTagHandler.default = endTagOther
+
     class InTableTextPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
+        __slots__ = ("originalPhase", "characterTokens")
+
+        def __init__(self, *args, **kwargs):
+            super(InTableTextPhase, self).__init__(*args, **kwargs)
             self.originalPhase = None
             self.characterTokens = []
 
@@ -1871,23 +1889,7 @@ def getPhases(debug):
 
     class InCaptionPhase(Phase):
         # http://www.whatwg.org/specs/web-apps/current-work/#in-caption
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                (("caption", "col", "colgroup", "tbody", "td", "tfoot", "th",
-                  "thead", "tr"), self.startTagTableElement)
-            ])
-            self.startTagHandler.default = self.startTagOther
-
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("caption", self.endTagCaption),
-                ("table", self.endTagTable),
-                (("body", "col", "colgroup", "html", "tbody", "td", "tfoot", "th",
-                  "thead", "tr"), self.endTagIgnore)
-            ])
-            self.endTagHandler.default = self.endTagOther
+        __slots__ = tuple()
 
         def ignoreEndTagCaption(self):
             return not self.tree.elementInScope("caption", variant="table")
@@ -1940,23 +1942,24 @@ def getPhases(debug):
         def endTagOther(self, token):
             return self.parser.phases["inBody"].processEndTag(token)
 
-    class InColumnGroupPhase(Phase):
-        # http://www.whatwg.org/specs/web-apps/current-work/#in-column
-
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("col", self.startTagCol)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            (("caption", "col", "colgroup", "tbody", "td", "tfoot", "th",
+              "thead", "tr"), startTagTableElement)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("colgroup", self.endTagColgroup),
-                ("col", self.endTagCol)
+        endTagHandler = _utils.MethodDispatcher([
+            ("caption", endTagCaption),
+            ("table", endTagTable),
+            (("body", "col", "colgroup", "html", "tbody", "td", "tfoot", "th",
+              "thead", "tr"), endTagIgnore)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InColumnGroupPhase(Phase):
+        # http://www.whatwg.org/specs/web-apps/current-work/#in-column
+        __slots__ = tuple()
 
         def ignoreEndTagColgroup(self):
             return self.tree.openElements[-1].name == "html"
@@ -2006,26 +2009,21 @@ def getPhases(debug):
             if not ignoreEndTag:
                 return token
 
-    class InTableBodyPhase(Phase):
-        # http://www.whatwg.org/specs/web-apps/current-work/#in-table0
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("tr", self.startTagTr),
-                (("td", "th"), self.startTagTableCell),
-                (("caption", "col", "colgroup", "tbody", "tfoot", "thead"),
-                 self.startTagTableOther)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            ("col", startTagCol)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                (("tbody", "tfoot", "thead"), self.endTagTableRowGroup),
-                ("table", self.endTagTable),
-                (("body", "caption", "col", "colgroup", "html", "td", "th",
-                  "tr"), self.endTagIgnore)
+        endTagHandler = _utils.MethodDispatcher([
+            ("colgroup", endTagColgroup),
+            ("col", endTagCol)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InTableBodyPhase(Phase):
+        # http://www.whatwg.org/specs/web-apps/current-work/#in-table0
+        __slots__ = tuple()
 
         # helper methods
         def clearStackToTableBodyContext(self):
@@ -2104,26 +2102,26 @@ def getPhases(debug):
         def endTagOther(self, token):
             return self.parser.phases["inTable"].processEndTag(token)
 
-    class InRowPhase(Phase):
-        # http://www.whatwg.org/specs/web-apps/current-work/#in-row
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                (("td", "th"), self.startTagTableCell),
-                (("caption", "col", "colgroup", "tbody", "tfoot", "thead",
-                  "tr"), self.startTagTableOther)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            ("tr", startTagTr),
+            (("td", "th"), startTagTableCell),
+            (("caption", "col", "colgroup", "tbody", "tfoot", "thead"),
+             startTagTableOther)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("tr", self.endTagTr),
-                ("table", self.endTagTable),
-                (("tbody", "tfoot", "thead"), self.endTagTableRowGroup),
-                (("body", "caption", "col", "colgroup", "html", "td", "th"),
-                 self.endTagIgnore)
+        endTagHandler = _utils.MethodDispatcher([
+            (("tbody", "tfoot", "thead"), endTagTableRowGroup),
+            ("table", endTagTable),
+            (("body", "caption", "col", "colgroup", "html", "td", "th",
+              "tr"), endTagIgnore)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InRowPhase(Phase):
+        # http://www.whatwg.org/specs/web-apps/current-work/#in-row
+        __slots__ = tuple()
 
         # helper methods (XXX unify this with other table helper methods)
         def clearStackToTableRowContext(self):
@@ -2193,23 +2191,26 @@ def getPhases(debug):
         def endTagOther(self, token):
             return self.parser.phases["inTable"].processEndTag(token)
 
-    class InCellPhase(Phase):
-        # http://www.whatwg.org/specs/web-apps/current-work/#in-cell
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                (("caption", "col", "colgroup", "tbody", "td", "tfoot", "th",
-                  "thead", "tr"), self.startTagTableOther)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            (("td", "th"), startTagTableCell),
+            (("caption", "col", "colgroup", "tbody", "tfoot", "thead",
+              "tr"), startTagTableOther)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                (("td", "th"), self.endTagTableCell),
-                (("body", "caption", "col", "colgroup", "html"), self.endTagIgnore),
-                (("table", "tbody", "tfoot", "thead", "tr"), self.endTagImply)
+        endTagHandler = _utils.MethodDispatcher([
+            ("tr", endTagTr),
+            ("table", endTagTable),
+            (("tbody", "tfoot", "thead"), endTagTableRowGroup),
+            (("body", "caption", "col", "colgroup", "html", "td", "th"),
+             endTagIgnore)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InCellPhase(Phase):
+        # http://www.whatwg.org/specs/web-apps/current-work/#in-cell
+        __slots__ = tuple()
 
         # helper
         def closeCell(self):
@@ -2269,26 +2270,22 @@ def getPhases(debug):
         def endTagOther(self, token):
             return self.parser.phases["inBody"].processEndTag(token)
 
-    class InSelectPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("option", self.startTagOption),
-                ("optgroup", self.startTagOptgroup),
-                ("select", self.startTagSelect),
-                (("input", "keygen", "textarea"), self.startTagInput),
-                ("script", self.startTagScript)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            (("caption", "col", "colgroup", "tbody", "td", "tfoot", "th",
+              "thead", "tr"), startTagTableOther)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("option", self.endTagOption),
-                ("optgroup", self.endTagOptgroup),
-                ("select", self.endTagSelect)
+        endTagHandler = _utils.MethodDispatcher([
+            (("td", "th"), endTagTableCell),
+            (("body", "caption", "col", "colgroup", "html"), endTagIgnore),
+            (("table", "tbody", "tfoot", "thead", "tr"), endTagImply)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InSelectPhase(Phase):
+        __slots__ = tuple()
 
         # http://www.whatwg.org/specs/web-apps/current-work/#in-select
         def processEOF(self):
@@ -2369,21 +2366,25 @@ def getPhases(debug):
             self.parser.parseError("unexpected-end-tag-in-select",
                                    {"name": token["name"]})
 
-    class InSelectInTablePhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                (("caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"),
-                 self.startTagTable)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            ("option", startTagOption),
+            ("optgroup", startTagOptgroup),
+            ("select", startTagSelect),
+            (("input", "keygen", "textarea"), startTagInput),
+            ("script", startTagScript)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                (("caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"),
-                 self.endTagTable)
+        endTagHandler = _utils.MethodDispatcher([
+            ("option", endTagOption),
+            ("optgroup", endTagOptgroup),
+            ("select", endTagSelect)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class InSelectInTablePhase(Phase):
+        __slots__ = tuple()
 
         def processEOF(self):
             self.parser.phases["inSelect"].processEOF()
@@ -2408,7 +2409,21 @@ def getPhases(debug):
         def endTagOther(self, token):
             return self.parser.phases["inSelect"].processEndTag(token)
 
+        startTagHandler = _utils.MethodDispatcher([
+            (("caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"),
+             startTagTable)
+        ])
+        startTagHandler.default = startTagOther
+
+        endTagHandler = _utils.MethodDispatcher([
+            (("caption", "table", "tbody", "tfoot", "thead", "tr", "td", "th"),
+             endTagTable)
+        ])
+        endTagHandler.default = endTagOther
+
     class InForeignContentPhase(Phase):
+        __slots__ = tuple()
+
         breakoutElements = frozenset(["b", "big", "blockquote", "body", "br",
                                       "center", "code", "dd", "div", "dl", "dt",
                                       "em", "embed", "h1", "h2", "h3",
@@ -2418,9 +2433,6 @@ def getPhases(debug):
                                       "span", "strong", "strike", "sub", "sup",
                                       "table", "tt", "u", "ul", "var"])
 
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
         def adjustSVGTagNames(self, token):
             replacements = {"altglyph": "altGlyph",
                             "altglyphdef": "altGlyphDef",
@@ -2524,16 +2536,7 @@ def getPhases(debug):
             return new_token
 
     class AfterBodyPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml)
-            ])
-            self.startTagHandler.default = self.startTagOther
-
-            self.endTagHandler = _utils.MethodDispatcher([("html", self.endTagHtml)])
-            self.endTagHandler.default = self.endTagOther
+        __slots__ = tuple()
 
         def processEOF(self):
             # Stop parsing
@@ -2570,23 +2573,17 @@ def getPhases(debug):
             self.parser.phase = self.parser.phases["inBody"]
             return token
 
-    class InFramesetPhase(Phase):
-        # http://www.whatwg.org/specs/web-apps/current-work/#in-frameset
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("frameset", self.startTagFrameset),
-                ("frame", self.startTagFrame),
-                ("noframes", self.startTagNoframes)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", startTagHtml)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("frameset", self.endTagFrameset)
-            ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler = _utils.MethodDispatcher([("html", endTagHtml)])
+        endTagHandler.default = endTagOther
+
+    class InFramesetPhase(Phase):
+        # http://www.whatwg.org/specs/web-apps/current-work/#in-frameset
+        __slots__ = tuple()
 
         def processEOF(self):
             if self.tree.openElements[-1].name != "html":
@@ -2627,21 +2624,22 @@ def getPhases(debug):
             self.parser.parseError("unexpected-end-tag-in-frameset",
                                    {"name": token["name"]})
 
-    class AfterFramesetPhase(Phase):
-        # http://www.whatwg.org/specs/web-apps/current-work/#after3
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("noframes", self.startTagNoframes)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            ("frameset", startTagFrameset),
+            ("frame", startTagFrame),
+            ("noframes", startTagNoframes)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
 
-            self.endTagHandler = _utils.MethodDispatcher([
-                ("html", self.endTagHtml)
+        endTagHandler = _utils.MethodDispatcher([
+            ("frameset", endTagFrameset)
         ])
-            self.endTagHandler.default = self.endTagOther
+        endTagHandler.default = endTagOther
+
+    class AfterFramesetPhase(Phase):
+        # http://www.whatwg.org/specs/web-apps/current-work/#after3
+        __slots__ = tuple()
 
         def processEOF(self):
             # Stop parsing
@@ -2664,14 +2662,19 @@ def getPhases(debug):
             self.parser.parseError("unexpected-end-tag-after-frameset",
                                    {"name": token["name"]})
 
-    class AfterAfterBodyPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", Phase.startTagHtml),
+            ("noframes", startTagNoframes)
+        ])
+        startTagHandler.default = startTagOther
 
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml)
+        endTagHandler = _utils.MethodDispatcher([
+            ("html", endTagHtml)
         ])
-            self.startTagHandler.default = self.startTagOther
+        endTagHandler.default = endTagOther
+
+    class AfterAfterBodyPhase(Phase):
+        __slots__ = tuple()
 
         def processEOF(self):
             pass
@@ -2702,15 +2705,13 @@ def getPhases(debug):
             self.parser.phase = self.parser.phases["inBody"]
             return token
 
-    class AfterAfterFramesetPhase(Phase):
-        def __init__(self, parser, tree):
-            Phase.__init__(self, parser, tree)
-
-            self.startTagHandler = _utils.MethodDispatcher([
-                ("html", self.startTagHtml),
-                ("noframes", self.startTagNoFrames)
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", startTagHtml)
         ])
-            self.startTagHandler.default = self.startTagOther
+        startTagHandler.default = startTagOther
+
+    class AfterAfterFramesetPhase(Phase):
+        __slots__ = tuple()
 
         def processEOF(self):
             pass
@@ -2737,6 +2738,13 @@ def getPhases(debug):
         def processEndTag(self, token):
             self.parser.parseError("expected-eof-but-got-end-tag",
                                    {"name": token["name"]})
+
+        startTagHandler = _utils.MethodDispatcher([
+            ("html", startTagHtml),
+            ("noframes", startTagNoFrames)
+        ])
+        startTagHandler.default = startTagOther
+
     # pylint:enable=unused-argument
 
     return {
@@ -2770,7 +2778,7 @@ def getPhases(debug):
 def adjust_attributes(token, replacements):
     needs_adjustment = viewkeys(token['data']) & viewkeys(replacements)
     if needs_adjustment:
-        token['data'] = OrderedDict((replacements.get(k, k), v)
+        token['data'] = type(token['data'])((replacements.get(k, k), v)
                                             for k, v in token['data'].items())
 
 

lib/html5lib/treebuilders/etree.py

@@ -5,6 +5,8 @@ from six import text_type
 
 import re
 
+from copy import copy
+
 from . import base
 from .. import _ihatexml
 from .. import constants
@@ -61,16 +63,17 @@ def getETreeBuilder(ElementTreeImplementation, fullTree=False):
             return self._element.attrib
 
         def _setAttributes(self, attributes):
-            # Delete existing attributes first
-            # XXX - there may be a better way to do this...
-            for key in list(self._element.attrib.keys()):
-                del self._element.attrib[key]
+            el_attrib = self._element.attrib
+            el_attrib.clear()
+            if attributes:
+                # calling .items _always_ allocates, and the above truthy check is cheaper than the
+                # allocation on average
                 for key, value in attributes.items():
                     if isinstance(key, tuple):
                         name = "{%s}%s" % (key[2], key[1])
                     else:
                         name = key
-                self._element.set(name, value)
+                    el_attrib[name] = value
 
         attributes = property(_getAttributes, _setAttributes)
 
@@ -129,8 +132,8 @@ def getETreeBuilder(ElementTreeImplementation, fullTree=False):
 
         def cloneNode(self):
             element = type(self)(self.name, self.namespace)
-            for name, value in self.attributes.items():
-                element.attributes[name] = value
+            if self._element.attrib:
+                element._element.attrib = copy(self._element.attrib)
             return element
 
         def reparentChildren(self, newParent):

lib/html5lib/treebuilders/etree_lxml.py

@@ -16,6 +16,11 @@ import warnings
 import re
 import sys
 
+try:
+    from collections.abc import MutableMapping
+except ImportError:
+    from collections import MutableMapping
+
 from . import base
 from ..constants import DataLossWarning
 from .. import constants
@@ -23,6 +28,7 @@ from . import etree as etree_builders
 from .. import _ihatexml
 
 import lxml.etree as etree
+from six import PY3, binary_type
 
 
 fullTree = True
@@ -44,7 +50,11 @@ class Document(object):
         self._childNodes = []
 
     def appendChild(self, element):
-        self._elementTree.getroot().addnext(element._element)
+        last = self._elementTree.getroot()
+        for last in self._elementTree.getroot().itersiblings():
+            pass
+
+        last.addnext(element._element)
 
     def _getChildNodes(self):
         return self._childNodes
@@ -185,26 +195,37 @@ class TreeBuilder(base.TreeBuilder):
         infosetFilter = self.infosetFilter = _ihatexml.InfosetFilter(preventDoubleDashComments=True)
         self.namespaceHTMLElements = namespaceHTMLElements
 
-        class Attributes(dict):
-            def __init__(self, element, value=None):
-                if value is None:
-                    value = {}
+        class Attributes(MutableMapping):
+            def __init__(self, element):
                 self._element = element
-                dict.__init__(self, value)  # pylint:disable=non-parent-init-called
-                for key, value in self.items():
+
+            def _coerceKey(self, key):
                 if isinstance(key, tuple):
                     name = "{%s}%s" % (key[2], infosetFilter.coerceAttribute(key[1]))
                 else:
                     name = infosetFilter.coerceAttribute(key)
-                    self._element._element.attrib[name] = value
+                return name
+
+            def __getitem__(self, key):
+                value = self._element._element.attrib[self._coerceKey(key)]
+                if not PY3 and isinstance(value, binary_type):
+                    value = value.decode("ascii")
+                return value
 
             def __setitem__(self, key, value):
-                dict.__setitem__(self, key, value)
-                if isinstance(key, tuple):
-                    name = "{%s}%s" % (key[2], infosetFilter.coerceAttribute(key[1]))
-                else:
-                    name = infosetFilter.coerceAttribute(key)
-                self._element._element.attrib[name] = value
+                self._element._element.attrib[self._coerceKey(key)] = value
+
+            def __delitem__(self, key):
+                del self._element._element.attrib[self._coerceKey(key)]
+
+            def __iter__(self):
+                return iter(self._element._element.attrib)
+
+            def __len__(self):
+                return len(self._element._element.attrib)
+
+            def clear(self):
+                return self._element._element.attrib.clear()
 
         class Element(builder.Element):
             def __init__(self, name, namespace):
@@ -225,8 +246,10 @@ class TreeBuilder(base.TreeBuilder):
             def _getAttributes(self):
                 return self._attributes
 
-            def _setAttributes(self, attributes):
-                self._attributes = Attributes(self, attributes)
+            def _setAttributes(self, value):
+                attributes = self.attributes
+                attributes.clear()
+                attributes.update(value)
 
             attributes = property(_getAttributes, _setAttributes)
 
@@ -234,8 +257,11 @@ class TreeBuilder(base.TreeBuilder):
                 data = infosetFilter.coerceCharacters(data)
                 builder.Element.insertText(self, data, insertBefore)
 
-            def appendChild(self, child):
-                builder.Element.appendChild(self, child)
+            def cloneNode(self):
+                element = type(self)(self.name, self.namespace)
+                if self._element.attrib:
+                    element._element.attrib.update(self._element.attrib)
+                return element
 
         class Comment(builder.Comment):
             def __init__(self, data):

lib/html5lib/treewalkers/etree_lxml.py

@@ -1,6 +1,8 @@
 from __future__ import absolute_import, division, unicode_literals
 from six import text_type
 
+from collections import OrderedDict
+
 from lxml import etree
 from ..treebuilders.etree import tag_regexp
 
@@ -163,7 +165,7 @@ class TreeWalker(base.NonRecursiveTreeWalker):
             else:
                 namespace = None
                 tag = ensure_str(node.tag)
-            attrs = {}
+            attrs = OrderedDict()
             for name, value in list(node.attrib.items()):
                 name = ensure_str(name)
                 value = ensure_str(value)