Skip to content

Commit c8e024b

Browse files
author
Marc DM
committed
fixed the unicode/str/six bug, but the test that proves it is in an alien file
1 parent 81b97d5 commit c8e024b

3 files changed

Lines changed: 45 additions & 28 deletions

File tree

‎html5lib/serializer/htmlserializer.py‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
from __future__ import absolute_import, division, unicode_literals
2-
from six import text_type
2+
from six import text_type, string_types
33

44
import gettext
55
_ = gettext.gettext
@@ -154,14 +154,14 @@ def __init__(self, **kwargs):
154154
self.strict = False
155155

156156
def encode(self, string):
157-
assert(isinstance(string, text_type))
157+
assert(isinstance(string, string_types))
158158
if self.encoding:
159159
return string.encode(self.encoding, unicode_encode_errors)
160160
else:
161161
return string
162162

163163
def encodeStrict(self, string):
164-
assert(isinstance(string, text_type))
164+
assert(isinstance(string, string_types))
165165
if self.encoding:
166166
return string.encode(self.encoding, "strict")
167167
else:

‎html5lib/tests/test_six_encoding.py‎

Lines changed: 17 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,17 @@
1+
2+
from html5lib import html5parser, treewalkers, serializer
3+
from nose.tools import eq_
4+
5+
def test_treewalker6():
6+
"""Str/Unicode mix. If str attrs added to tree"""
7+
8+
text = '<a href="http://example.com">Example</a>'
9+
end_text = '<a href="http://example.com" class="test123">Example</a>'
10+
parser = html5parser.HTMLParser()
11+
walker = treewalkers.getTreeWalker('etree')
12+
serializr = serializer.HTMLSerializer(quote_attr_values=True)
13+
domtree = parser.parseFragment(text)
14+
15+
# at this point domtree should be a DOCUMENT_FRAGMENT
16+
domtree[0].set('class', 'test123')
17+
eq_(end_text, serializr.render(walker(domtree)))

‎html5lib/treewalkers/_base.py‎

Lines changed: 25 additions & 25 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
from __future__ import absolute_import, division, unicode_literals
2-
from six import text_type
2+
from six import text_type, string_types
33

44
import gettext
55
_ = gettext.gettext
@@ -19,43 +19,43 @@ def error(self, msg):
1919
return {"type": "SerializeError", "data": msg}
2020

2121
def emptyTag(self, namespace, name, attrs, hasChildren=False):
22-
assert namespace is None or isinstance(namespace, text_type), type(namespace)
23-
assert isinstance(name, text_type), type(name)
24-
assert all((namespace is None or isinstance(namespace, text_type)) and
25-
isinstance(name, text_type) and
26-
isinstance(value, text_type)
22+
assert namespace is None or isinstance(namespace, string_types), type(namespace)
23+
assert isinstance(name, string_types), type(name)
24+
assert all((namespace is None or isinstance(namespace, string_types)) and
25+
isinstance(name, string_types) and
26+
isinstance(value, string_types)
2727
for (namespace, name), value in attrs.items())
2828

29-
yield {"type": "EmptyTag", "name": name,
30-
"namespace": namespace,
29+
yield {"type": "EmptyTag", "name": text_type(name),
30+
"namespace": text_type(namespace),
3131
"data": attrs}
3232
if hasChildren:
3333
yield self.error(_("Void element has children"))
3434

3535
def startTag(self, namespace, name, attrs):
36-
assert namespace is None or isinstance(namespace, text_type), type(namespace)
37-
assert isinstance(name, text_type), type(name)
38-
assert all((namespace is None or isinstance(namespace, text_type)) and
39-
isinstance(name, text_type) and
40-
isinstance(value, text_type)
36+
assert namespace is None or isinstance(namespace, string_types), type(namespace)
37+
assert isinstance(name, string_types), type(name)
38+
assert all((namespace is None or isinstance(namespace, string_types)) and
39+
isinstance(name, string_types) and
40+
isinstance(value, string_types)
4141
for (namespace, name), value in attrs.items())
4242

4343
return {"type": "StartTag",
44-
"name": name,
45-
"namespace": namespace,
44+
"name": text_type(name),
45+
"namespace": text_type(namespace),
4646
"data": attrs}
4747

4848
def endTag(self, namespace, name):
49-
assert namespace is None or isinstance(namespace, text_type), type(namespace)
50-
assert isinstance(name, text_type), type(namespace)
49+
assert namespace is None or isinstance(namespace, string_types), type(namespace)
50+
assert isinstance(name, string_types), type(namespace)
5151

5252
return {"type": "EndTag",
53-
"name": name,
54-
"namespace": namespace,
53+
"name": text_type(name),
54+
"namespace": text_type(namespace),
5555
"data": {}}
5656

5757
def text(self, data):
58-
assert isinstance(data, text_type), type(data)
58+
assert isinstance(data, string_types), type(data)
5959

6060
data = data
6161
middle = data.lstrip(spaceCharacters)
@@ -71,14 +71,14 @@ def text(self, data):
7171
yield {"type": "SpaceCharacters", "data": right}
7272

7373
def comment(self, data):
74-
assert isinstance(data, text_type), type(data)
74+
assert isinstance(data, string_types), type(data)
7575

7676
return {"type": "Comment", "data": data}
7777

7878
def doctype(self, name, publicId=None, systemId=None, correct=True):
79-
assert name is None or isinstance(name, text_type), type(name)
80-
assert publicId is None or isinstance(publicId, text_type), type(publicId)
81-
assert systemId is None or isinstance(systemId, text_type), type(systemId)
79+
assert name is None or isinstance(name, string_types), type(name)
80+
assert publicId is None or isinstance(publicId, string_types), type(publicId)
81+
assert systemId is None or isinstance(systemId, string_types), type(systemId)
8282

8383
return {"type": "Doctype",
8484
"name": name if name is not None else "",
@@ -87,7 +87,7 @@ def doctype(self, name, publicId=None, systemId=None, correct=True):
8787
"correct": correct}
8888

8989
def entity(self, name):
90-
assert isinstance(name, text_type), type(name)
90+
assert isinstance(name, string_types), type(name)
9191

9292
return {"type": "Entity", "name": name}
9393

0 commit comments

Comments
 (0)