Fix #37: Preserve order of attributes on serialization.

This doesn't do anything about the fact that none of our
treebuilders preserve attribute order: it merely avoids the
serializer reordering them from the order it receives them in.

This also provides a filter to obtain the previous behaviour of
lexicographical ordering, used by the serializer tests to ensure
determinism of the output.

Author: gsnedders

.travis.yml

@@ -28,6 +28,7 @@ before_install:
 install:
   - pip install -r requirements.txt -r requirements-test.txt --use-mirrors
   - if [[ $USE_OPTIONAL == "true" ]]; then pip install -r requirements-optional.txt --use-mirrors; fi
+  - if [[ $TRAVIS_PYTHON_VERSION == "2.6" && $USE_OPTIONAL == "true" ]]; then pip install -r requirements-optional-2.6.txt --use-mirrors; fi
   - if [[ $TRAVIS_PYTHON_VERSION != "pypy" && $USE_OPTIONAL == "true" ]]; then pip install -r requirements-optional-cpython.txt --use-mirrors; fi
   - if [[ $FLAKE == "true" ]]; then pip install --use-mirrors flake8; fi
 

README.rst

@@ -63,6 +63,10 @@ format:
 
 More documentation is available at http://html5lib.readthedocs.org/.
 
+- ``ordereddict`` can be used under Python 2.6
+  (``collections.OrderedDict`` is used instead on later versions) to
+  serialize attributes in alphabetical order.
+
 
 Installation
 ------------
@@ -106,7 +110,8 @@ Tests
 -----
 
 Unit tests require the ``nose`` library and can be run using the
-``nosetests`` command in the root directory. All should pass.
+``nosetests`` command in the root directory; ``ordereddict`` is
+required under Python 2.6. All should pass.
 
 Test data are contained in a separate `html5lib-tests
 <https://github.com/html5lib/html5lib-tests>`_ repository and included
@@ -115,8 +120,6 @@ as a submodule, thus for git checkouts they must be initialized::
   $ git submodule init
   $ git submodule update
 
-This is unneeded for release tarballs.
-
 If you have all compatible Python implementations available on your
 system, you can run tests on all of them using the ``tox`` utility,
 which can be found on PyPI.

html5lib/filters/alphabeticalattributes.py

@@ -0,0 +1,20 @@
+from __future__ import absolute_import, division, unicode_literals
+
+from . import _base
+
+try:
+    from collections import OrderedDict
+except ImportError:
+    from ordereddict import OrderedDict
+
+
+class Filter(_base.Filter):
+    def __iter__(self):
+        for token in _base.Filter.__iter__(self):
+            if token["type"] in ("StartTag", "EmptyTag"):
+                attrs = OrderedDict()
+                for name, value in sorted(token["data"].items(),
+                                          key=lambda x: x[0]):
+                    attrs[name] = value
+                token["data"] = attrs
+            yield token

html5lib/serializer/htmlserializer.py

@@ -221,7 +221,7 @@ def serialize(self, treewalker, encoding=None):
                     in_cdata = True
                 elif in_cdata:
                     self.serializeError(_("Unexpected child element of a CDATA element"))
-                for (attr_namespace, attr_name), attr_value in sorted(token["data"].items()):
+                for (attr_namespace, attr_name), attr_value in token["data"].items():
                     # TODO: Add namespace support here
                     k = attr_name
                     v = attr_value

html5lib/tests/test_serializer.py

@@ -5,13 +5,20 @@
 
 from .support import get_data_files
 
+try:
+    from collections import OrderedDict
+except ImportError:
+    # Python 2.6 support
+    from ordereddict import OrderedDict
+
 try:
     unittest.TestCase.assertEqual
 except AttributeError:
     unittest.TestCase.assertEqual = unittest.TestCase.assertEquals
 
 import html5lib
 from html5lib import serializer, constants
+from html5lib.filters.alphabeticalattributes import Filter as AlphabeticalAttributesFilter
 from html5lib.treewalkers._base import TreeWalker
 
 optionals_loaded = []
@@ -81,7 +88,8 @@ def _convertAttrib(self, attribs):
 
 def serialize_html(input, options):
     options = dict([(str(k), v) for k, v in options.items()])
-    return serializer.HTMLSerializer(**options).render(JsonWalker(input), options.get("encoding", None))
+    stream = AlphabeticalAttributesFilter(JsonWalker(input))
+    return serializer.HTMLSerializer(**options).render(stream, options.get("encoding", None))
 
 
 def runSerializerTest(input, expected, options):

requirements-optional-2.6.txt

@@ -0,0 +1,3 @@
+# Can be used to force attributes to be serialized in alphabetical
+# order.
+ordereddict

requirements-test.txt

@@ -1 +1,2 @@
 nose
+ordereddict # Python 2.6

tox.ini

@@ -20,3 +20,14 @@ deps =
   Genshi
   nose
   six
+
+[testenv:py26]
+basepython = python2.6
+deps =
+  charade
+  datrie
+  Genshi
+  lxml
+  nose
+  six
+  ordereddict