Remove six dependency from html5lib (#618)

This way lies madness, but at least we don't have a six dependency
anymore.

The way this work is that we vendored html5lib 1.1, but then this
applies a 01_html5lib_six.patch to that which changes imports from six
to import from bleach.six_shim.

This updates the vendor management code and vendorverify to install
html5lib 1.1 and then apply the patch and then compare with what's in
the tree. If we end up applying further patches in the future, we can
use this model to do that.

Author: willkg

bleach/_vendor/html5lib/_inputstream.py

@@ -1,7 +1,7 @@
 from __future__ import absolute_import, division, unicode_literals
 
-from six import text_type
-from six.moves import http_client, urllib
+from bleach.six_shim import text_type
+from bleach.six_shim import http_client, urllib
 
 import codecs
 import re

bleach/_vendor/html5lib/_tokenizer.py

@@ -1,6 +1,6 @@
 from __future__ import absolute_import, division, unicode_literals
 
-from six import unichr as chr
+from bleach.six_shim import unichr as chr
 
 from collections import deque, OrderedDict
 from sys import version_info

bleach/_vendor/html5lib/_trie/py.py

@@ -1,5 +1,5 @@
 from __future__ import absolute_import, division, unicode_literals
-from six import text_type
+from bleach.six_shim import text_type
 
 from bisect import bisect_left
 

bleach/_vendor/html5lib/_utils.py

@@ -7,7 +7,7 @@
 except ImportError:
     from collections import Mapping
 
-from six import text_type, PY3
+from bleach.six_shim import text_type, PY3
 
 if PY3:
     import xml.etree.ElementTree as default_etree

bleach/_vendor/html5lib/filters/lint.py

@@ -1,6 +1,6 @@
 from __future__ import absolute_import, division, unicode_literals
 
-from six import text_type
+from bleach.six_shim import text_type
 
 from . import base
 from ..constants import namespaces, voidElements

bleach/_vendor/html5lib/filters/sanitizer.py

@@ -12,7 +12,7 @@
 import warnings
 from xml.sax.saxutils import escape, unescape
 
-from six.moves import urllib_parse as urlparse
+from bleach.six_shim import urllib_parse as urlparse
 
 from . import base
 from ..constants import namespaces, prefixes

bleach/_vendor/html5lib/html5parser.py

@@ -1,5 +1,5 @@
 from __future__ import absolute_import, division, unicode_literals
-from six import with_metaclass, viewkeys
+from bleach.six_shim import viewkeys
 
 import types
 
@@ -423,7 +423,7 @@ def getMetaclass(use_metaclass, metaclass_func):
             return type
 
     # pylint:disable=unused-argument
-    class Phase(with_metaclass(getMetaclass(debug, log))):
+    class Phase(metaclass=getMetaclass(debug, log)):
         """Base class for helper object that implements each phase of processing
         """
         __slots__ = ("parser", "tree", "__startTagCache", "__endTagCache")

bleach/_vendor/html5lib/serializer.py

@@ -1,5 +1,5 @@
 from __future__ import absolute_import, division, unicode_literals
-from six import text_type
+from bleach.six_shim import text_type
 
 import re
 

bleach/_vendor/html5lib/treebuilders/base.py

@@ -1,5 +1,5 @@
 from __future__ import absolute_import, division, unicode_literals
-from six import text_type
+from bleach.six_shim import text_type
 
 from ..constants import scopingElements, tableInsertModeElements, namespaces
 

bleach/_vendor/html5lib/treebuilders/etree.py

@@ -1,7 +1,7 @@
 from __future__ import absolute_import, division, unicode_literals
 # pylint:disable=protected-access
 
-from six import text_type
+from bleach.six_shim import text_type
 
 import re
 

bleach/_vendor/html5lib/treebuilders/etree_lxml.py

@@ -28,7 +28,7 @@
 from .. import _ihatexml
 
 import lxml.etree as etree
-from six import PY3, binary_type
+from bleach.six_shim import PY3, binary_type
 
 
 fullTree = True

bleach/_vendor/html5lib/treewalkers/etree.py

@@ -3,7 +3,7 @@
 from collections import OrderedDict
 import re
 
-from six import string_types
+from bleach.six_shim import string_types
 
 from . import base
 from .._utils import moduleFactoryFactory

bleach/_vendor/html5lib/treewalkers/etree_lxml.py

@@ -1,5 +1,5 @@
 from __future__ import absolute_import, division, unicode_literals
-from six import text_type
+from bleach.six_shim import text_type
 
 from collections import OrderedDict
 

bleach/_vendor/vendor_install.sh

@@ -7,8 +7,12 @@ set -o pipefail
 BLEACH_VENDOR_DIR=${BLEACH_VENDOR_DIR:-"."}
 DEST=${DEST:-"."}
 
+# Install with no dependencies
 pip install --no-binary all --no-compile --no-deps -r "${BLEACH_VENDOR_DIR}/vendor.txt" --target "${DEST}"
 
+# Apply patches
+(cd "${DEST}" && patch -p2 < 01_html5lib_six.patch)
+
 # install Python 3.6.14 urllib.urlparse for #536
 curl --proto '=https' --tlsv1.2 -o "${DEST}/parse.py" https://raw.githubusercontent.com/python/cpython/v3.6.14/Lib/urllib/parse.py
 (cd "${DEST}" && sha256sum parse.py > parse.py.SHA256SUM)

scripts/vendor_verify.sh

@@ -17,6 +17,9 @@ mkdir "${DEST}"
 # Get versions of pip and python
 pip --version
 
+# Copy patch files to dest directory
+cp bleach/_vendor/*.patch "${DEST}"
+
 # Install vendored dependencies into temp directory
 BLEACH_VENDOR_DIR=bleach/_vendor DEST="${DEST}" bleach/_vendor/vendor_install.sh