diff --git a/CHANGES.rst b/CHANGES.rst
index 62a6a233..1b557816 100644
--- a/CHANGES.rst
+++ b/CHANGES.rst
@@ -33,6 +33,13 @@ Released on XXX
* **Use scripting disabled by default (as we don't implement
scripting).**
+* **Fix #11, avoiding the XSS bug potentially caused by serializer
+ allowing attribute values to be escaped out of in old browser versions,
+ changing the quote_attr_values option on serializer to take one of
+ three values, "always" (the old True value), "legacy" (the new option,
+ and the new default), and "spec" (the old False value, and the old
+ default).**
+
0.9999999/1.0b8
~~~~~~~~~~~~~~~
diff --git a/html5lib/filters/lint.py b/html5lib/filters/lint.py
index 3ec63d72..af231d8e 100644
--- a/html5lib/filters/lint.py
+++ b/html5lib/filters/lint.py
@@ -10,6 +10,10 @@
class Filter(_base.Filter):
+ def __init__(self, source, require_matching_tags=True):
+ super(Filter, self).__init__(source)
+ self.require_matching_tags = require_matching_tags
+
def __iter__(self):
open_elements = []
for token in _base.Filter.__iter__(self):
@@ -26,7 +30,7 @@ def __iter__(self):
assert type == "EmptyTag"
else:
assert type == "StartTag"
- if type == "StartTag":
+ if type == "StartTag" and self.require_matching_tags:
open_elements.append((namespace, name))
for (namespace, name), value in token["data"].items():
assert namespace is None or isinstance(namespace, text_type)
@@ -44,7 +48,7 @@ def __iter__(self):
assert name != ""
if (not namespace or namespace == namespaces["html"]) and name in voidElements:
assert False, "Void element reported as EndTag token: %(tag)s" % {"tag": name}
- else:
+ elif self.require_matching_tags:
start = open_elements.pop()
assert start == (namespace, name)
diff --git a/html5lib/serializer/htmlserializer.py b/html5lib/serializer/htmlserializer.py
index b87d9a75..afe2e0e2 100644
--- a/html5lib/serializer/htmlserializer.py
+++ b/html5lib/serializer/htmlserializer.py
@@ -1,10 +1,7 @@
from __future__ import absolute_import, division, unicode_literals
from six import text_type
-try:
- from functools import reduce
-except ImportError:
- pass
+import re
from ..constants import voidElements, booleanAttributes, spaceCharacters
from ..constants import rcdataElements, entities, xmlEntities
@@ -13,6 +10,17 @@
spaceCharacters = "".join(spaceCharacters)
+quoteAttributeSpecChars = spaceCharacters + "\"'=<>`"
+quoteAttributeSpec = re.compile("[" + quoteAttributeSpecChars + "]")
+quoteAttributeLegacy = re.compile("[" + quoteAttributeSpecChars +
+ "\x00\x01\x02\x03\x04\x05\x06\x07\x08\t\n"
+ "\x0b\x0c\r\x0e\x0f\x10\x11\x12\x13\x14\x15"
+ "\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f"
+ "\x20\x2f\x60\xa0\u1680\u180e\u180f\u2000"
+ "\u2001\u2002\u2003\u2004\u2005\u2006\u2007"
+ "\u2008\u2009\u200a\u2028\u2029\u202f\u205f"
+ "\u3000]")
+
try:
from codecs import register_error, xmlcharrefreplace_errors
except ImportError:
@@ -73,7 +81,7 @@ def htmlentityreplace_errors(exc):
class HTMLSerializer(object):
# attribute quoting options
- quote_attr_values = False
+ quote_attr_values = "legacy" # be secure by default
quote_char = '"'
use_best_quote_char = True
@@ -109,9 +117,9 @@ def __init__(self, **kwargs):
inject_meta_charset=True|False
Whether it insert a meta element to define the character set of the
document.
- quote_attr_values=True|False
+ quote_attr_values="legacy"|"spec"|"always"
Whether to quote attribute values that don't require quoting
- per HTML5 parsing rules.
+ per legacy browser behaviour, when required by the standard, or always.
quote_char=u'"'|u"'"
Use given quote character for attribute quoting. Default is to
use double quote unless attribute value contains a double quote,
@@ -240,11 +248,15 @@ def serialize(self, treewalker, encoding=None):
(k not in booleanAttributes.get(name, tuple()) and
k not in booleanAttributes.get("", tuple())):
yield self.encodeStrict("=")
- if self.quote_attr_values or not v:
+ if self.quote_attr_values == "always" or len(v) == 0:
quote_attr = True
+ elif self.quote_attr_values == "spec":
+ quote_attr = quoteAttributeSpec.search(v) is not None
+ elif self.quote_attr_values == "legacy":
+ quote_attr = quoteAttributeLegacy.search(v) is not None
else:
- quote_attr = reduce(lambda x, y: x or (y in v),
- spaceCharacters + ">\"'=", False)
+ raise ValueError("quote_attr_values must be one of: "
+ "'always', 'spec', or 'legacy'")
v = v.replace("&", "&")
if self.escape_lt_in_attrs:
v = v.replace("<", "<")
diff --git a/html5lib/tests/serializer-testdata/core.test b/html5lib/tests/serializer-testdata/core.test
new file mode 100644
index 00000000..70828d0d
--- /dev/null
+++ b/html5lib/tests/serializer-testdata/core.test
@@ -0,0 +1,395 @@
+{
+ "tests": [
+ {
+ "expected": [
+ "<span title='test \"with\" &quot;'>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "test \"with\" ""
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value escaping"
+ },
+ {
+ "expected": [
+ "<span title=foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value non-quoting"
+ },
+ {
+ "expected": [
+ "<span title=\"foo<bar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo<bar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value non-quoting (with <)"
+ },
+ {
+ "expected": [
+ "<span title=\"foo=bar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo=bar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with =)"
+ },
+ {
+ "expected": [
+ "<span title=\"foo>bar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo>bar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with >)"
+ },
+ {
+ "expected": [
+ "<span title='foo\"bar'>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo\"bar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with \")"
+ },
+ {
+ "expected": [
+ "<span title=\"foo'bar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo'bar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with ')"
+ },
+ {
+ "expected": [
+ "<span title=\"foo'bar"baz\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo'bar\"baz"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with both \" and ')"
+ },
+ {
+ "expected": [
+ "<span title=\"foo bar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo bar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with space)"
+ },
+ {
+ "expected": [
+ "<span title=\"foo\tbar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo\tbar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with tab)"
+ },
+ {
+ "expected": [
+ "<span title=\"foo\nbar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo\nbar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with LF)"
+ },
+ {
+ "expected": [
+ "<span title=\"foo\rbar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo\rbar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with CR)"
+ },
+ {
+ "expected": [
+ "<span title=\"foo\u000bbar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo\u000bbar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value non-quoting (with linetab)"
+ },
+ {
+ "expected": [
+ "<span title=\"foo\fbar\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "foo\fbar"
+ }
+ ]
+ ]
+ ],
+ "description": "proper attribute value quoting (with form feed)"
+ },
+ {
+ "expected": [
+ "<img>"
+ ],
+ "input": [
+ [
+ "EmptyTag",
+ "img",
+ {}
+ ]
+ ],
+ "description": "void element (as EmptyTag token)"
+ },
+ {
+ "expected": [
+ "<!DOCTYPE foo>"
+ ],
+ "input": [
+ [
+ "Doctype",
+ "foo"
+ ]
+ ],
+ "description": "doctype in error"
+ },
+ {
+ "expected": [
+ "a<b>c&d"
+ ],
+ "input": [
+ [
+ "Characters",
+ "a<b>c&d"
+ ]
+ ],
+ "description": "character data",
+ "options": {
+ "encoding": "utf-8"
+ }
+ },
+ {
+ "expected": [
+ "<script>a<b>c&d"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "script",
+ {}
+ ],
+ [
+ "Characters",
+ "a<b>c&d"
+ ]
+ ],
+ "description": "rcdata"
+ },
+ {
+ "expected": [
+ "<!DOCTYPE HTML>"
+ ],
+ "input": [
+ [
+ "Doctype",
+ "HTML"
+ ]
+ ],
+ "description": "doctype"
+ },
+ {
+ "expected": [
+ "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\" \"http://www.w3.org/TR/html4/strict.dtd\">"
+ ],
+ "input": [
+ [
+ "Doctype",
+ "HTML",
+ "-//W3C//DTD HTML 4.01//EN",
+ "http://www.w3.org/TR/html4/strict.dtd"
+ ]
+ ],
+ "description": "HTML 4.01 DOCTYPE"
+ },
+ {
+ "expected": [
+ "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\">"
+ ],
+ "input": [
+ [
+ "Doctype",
+ "HTML",
+ "-//W3C//DTD HTML 4.01//EN"
+ ]
+ ],
+ "description": "HTML 4.01 DOCTYPE without system identifer"
+ },
+ {
+ "expected": [
+ "<!DOCTYPE html SYSTEM \"http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd\">"
+ ],
+ "input": [
+ [
+ "Doctype",
+ "html",
+ "",
+ "http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd"
+ ]
+ ],
+ "description": "IBM DOCTYPE without public identifer"
+ }
+ ]
+}
diff --git a/html5lib/tests/serializer-testdata/injectmeta.test b/html5lib/tests/serializer-testdata/injectmeta.test
new file mode 100644
index 00000000..399590c3
--- /dev/null
+++ b/html5lib/tests/serializer-testdata/injectmeta.test
@@ -0,0 +1,350 @@
+{
+ "tests": [
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "no encoding",
+ "options": {
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta charset=utf-8>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "empytag head",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta charset=utf-8><title>foo</title>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "title",
+ {}
+ ],
+ [
+ "Characters",
+ "foo"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "title"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head w/title",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta charset=utf-8>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "charset",
+ "value": "ascii"
+ }
+ ]
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head w/meta-charset",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta charset=utf-8><meta charset=utf-8>",
+ "<head><meta charset=utf-8><meta charset=ascii>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "charset",
+ "value": "ascii"
+ }
+ ]
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "charset",
+ "value": "ascii"
+ }
+ ]
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head w/ two meta-charset",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta charset=utf-8><meta content=noindex name=robots>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "name",
+ "value": "robots"
+ },
+ {
+ "namespace": null,
+ "name": "content",
+ "value": "noindex"
+ }
+ ]
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head w/robots",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta content=noindex name=robots><meta charset=utf-8>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "name",
+ "value": "robots"
+ },
+ {
+ "namespace": null,
+ "name": "content",
+ "value": "noindex"
+ }
+ ]
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "charset",
+ "value": "ascii"
+ }
+ ]
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head w/robots & charset",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta content=\"text/html; charset=utf-8\" http-equiv=content-type>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "http-equiv",
+ "value": "content-type"
+ },
+ {
+ "namespace": null,
+ "name": "content",
+ "value": "text/html; charset=ascii"
+ }
+ ]
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head w/ charset in http-equiv content-type",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ },
+ {
+ "expected": [
+ "<meta content=noindex name=robots><meta content=\"text/html; charset=utf-8\" http-equiv=content-type>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "name",
+ "value": "robots"
+ },
+ {
+ "namespace": null,
+ "name": "content",
+ "value": "noindex"
+ }
+ ]
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ [
+ {
+ "namespace": null,
+ "name": "http-equiv",
+ "value": "content-type"
+ },
+ {
+ "namespace": null,
+ "name": "content",
+ "value": "text/html; charset=ascii"
+ }
+ ]
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head w/robots & charset in http-equiv content-type",
+ "options": {
+ "encoding": "utf-8",
+ "inject_meta_charset": true
+ }
+ }
+ ]
+}
\ No newline at end of file
diff --git a/html5lib/tests/serializer-testdata/optionaltags.test b/html5lib/tests/serializer-testdata/optionaltags.test
new file mode 100644
index 00000000..e67725ca
--- /dev/null
+++ b/html5lib/tests/serializer-testdata/optionaltags.test
@@ -0,0 +1,3254 @@
+{
+ "tests": [
+ {
+ "expected": [
+ "<html lang=en>foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "html",
+ [
+ {
+ "namespace": null,
+ "name": "lang",
+ "value": "en"
+ }
+ ]
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "html start-tag followed by text, with attributes"
+ },
+ {
+ "expected": [
+ "<html><!--foo-->"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "html",
+ {}
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "html start-tag followed by comment"
+ },
+ {
+ "expected": [
+ "<html> foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "html",
+ {}
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "html start-tag followed by space character"
+ },
+ {
+ "expected": [
+ "foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "html",
+ {}
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "html start-tag followed by text"
+ },
+ {
+ "expected": [
+ "<foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "html",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "html start-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "html",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "html start-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "html",
+ {}
+ ]
+ ],
+ "description": "html start-tag at EOF (shouldn't ever happen?!)"
+ },
+ {
+ "expected": [
+ "</html><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "html"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "html end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</html> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "html"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "html end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "html"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "html end-tag followed by text"
+ },
+ {
+ "expected": [
+ "<foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "html"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "html end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "html"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "html end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "html"
+ ]
+ ],
+ "description": "html end-tag at EOF"
+ },
+ {
+ "expected": [
+ "<head><!--foo-->"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "head start-tag followed by comment"
+ },
+ {
+ "expected": [
+ "<head> foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "head start-tag followed by space character"
+ },
+ {
+ "expected": [
+ "<head>foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "head start-tag followed by text"
+ },
+ {
+ "expected": [
+ "<foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "head start-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<head></foo>",
+ "</foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "head start-tag followed by end-tag (shouldn't ever happen?!)"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "empty head element"
+ },
+ {
+ "expected": [
+ "<meta>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "meta",
+ {}
+ ]
+ ],
+ "description": "head start-tag followed by empty-tag"
+ },
+ {
+ "expected": [
+ "<head>",
+ ""
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "head",
+ {}
+ ]
+ ],
+ "description": "head start-tag at EOF (shouldn't ever happen?!)"
+ },
+ {
+ "expected": [
+ "</head><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "head end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</head> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "head end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "head end-tag followed by text"
+ },
+ {
+ "expected": [
+ "<foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "head end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "head end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "head"
+ ]
+ ],
+ "description": "head end-tag at EOF"
+ },
+ {
+ "expected": [
+ "<body><!--foo-->"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "body",
+ {}
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "body start-tag followed by comment"
+ },
+ {
+ "expected": [
+ "<body> foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "body",
+ {}
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "body start-tag followed by space character"
+ },
+ {
+ "expected": [
+ "foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "body",
+ {}
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "body start-tag followed by text"
+ },
+ {
+ "expected": [
+ "<foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "body",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "body start-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "body",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "body start-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "body",
+ {}
+ ]
+ ],
+ "description": "body start-tag at EOF (shouldn't ever happen?!)"
+ },
+ {
+ "expected": [
+ "</body><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "body"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "body end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</body> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "body"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "body end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "body"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "body end-tag followed by text"
+ },
+ {
+ "expected": [
+ "<foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "body"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "body end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "body"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "body end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "body"
+ ]
+ ],
+ "description": "body end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</li><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "li"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "li end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</li> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "li"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "li end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</li>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "li"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "li end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</li><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "li"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "li end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<li>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "li"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "li",
+ {}
+ ]
+ ],
+ "description": "li end-tag followed by li start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "li"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "li end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "li"
+ ]
+ ],
+ "description": "li end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</dt><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "dt end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</dt> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "dt end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</dt>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "dt end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</dt><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "dt end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<dt>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt",
+ {}
+ ]
+ ],
+ "description": "dt end-tag followed by dt start-tag"
+ },
+ {
+ "expected": [
+ "<dd>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd",
+ {}
+ ]
+ ],
+ "description": "dt end-tag followed by dd start-tag"
+ },
+ {
+ "expected": [
+ "</dt></foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "dt end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ "</dt>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt"
+ ]
+ ],
+ "description": "dt end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</dd><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "dd end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</dd> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "dd end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</dd>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "dd end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</dd><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "dd end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<dd>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd",
+ {}
+ ]
+ ],
+ "description": "dd end-tag followed by dd start-tag"
+ },
+ {
+ "expected": [
+ "<dt>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "dt",
+ {}
+ ]
+ ],
+ "description": "dd end-tag followed by dt start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "dd end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "dd"
+ ]
+ ],
+ "description": "dd end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</p><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "p end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</p> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "p end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</p>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "p end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</p><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<address>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "address",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by address start-tag"
+ },
+ {
+ "expected": [
+ "<article>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "article",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by article start-tag"
+ },
+ {
+ "expected": [
+ "<aside>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "aside",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by aside start-tag"
+ },
+ {
+ "expected": [
+ "<blockquote>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "blockquote",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by blockquote start-tag"
+ },
+ {
+ "expected": [
+ "<datagrid>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "datagrid",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by datagrid start-tag"
+ },
+ {
+ "expected": [
+ "<dialog>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "dialog",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by dialog start-tag"
+ },
+ {
+ "expected": [
+ "<dir>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "dir",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by dir start-tag"
+ },
+ {
+ "expected": [
+ "<div>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by div start-tag"
+ },
+ {
+ "expected": [
+ "<dl>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "dl",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by dl start-tag"
+ },
+ {
+ "expected": [
+ "<fieldset>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "fieldset",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by fieldset start-tag"
+ },
+ {
+ "expected": [
+ "<footer>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "footer",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by footer start-tag"
+ },
+ {
+ "expected": [
+ "<form>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "form",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by form start-tag"
+ },
+ {
+ "expected": [
+ "<h1>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "h1",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by h1 start-tag"
+ },
+ {
+ "expected": [
+ "<h2>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "h2",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by h2 start-tag"
+ },
+ {
+ "expected": [
+ "<h3>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "h3",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by h3 start-tag"
+ },
+ {
+ "expected": [
+ "<h4>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "h4",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by h4 start-tag"
+ },
+ {
+ "expected": [
+ "<h5>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "h5",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by h5 start-tag"
+ },
+ {
+ "expected": [
+ "<h6>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "h6",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by h6 start-tag"
+ },
+ {
+ "expected": [
+ "<header>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "header",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by header start-tag"
+ },
+ {
+ "expected": [
+ "<hr>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "EmptyTag",
+ "hr",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by hr empty-tag"
+ },
+ {
+ "expected": [
+ "<menu>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "menu",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by menu start-tag"
+ },
+ {
+ "expected": [
+ "<nav>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "nav",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by nav start-tag"
+ },
+ {
+ "expected": [
+ "<ol>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "ol",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by ol start-tag"
+ },
+ {
+ "expected": [
+ "<p>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "p",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by p start-tag"
+ },
+ {
+ "expected": [
+ "<pre>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "pre",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by pre start-tag"
+ },
+ {
+ "expected": [
+ "<section>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "section",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by section start-tag"
+ },
+ {
+ "expected": [
+ "<table>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "table",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by table start-tag"
+ },
+ {
+ "expected": [
+ "<ul>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "ul",
+ {}
+ ]
+ ],
+ "description": "p end-tag followed by ul start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "p end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "p"
+ ]
+ ],
+ "description": "p end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</optgroup><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "optgroup end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</optgroup> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "optgroup end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</optgroup>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "optgroup end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</optgroup><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "optgroup end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<optgroup>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup",
+ {}
+ ]
+ ],
+ "description": "optgroup end-tag followed by optgroup start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "optgroup end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup"
+ ]
+ ],
+ "description": "optgroup end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</option><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "option end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</option> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "option end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</option>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "option end-tag followed by text"
+ },
+ {
+ "expected": [
+ "<optgroup>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "optgroup",
+ {}
+ ]
+ ],
+ "description": "option end-tag followed by optgroup start-tag"
+ },
+ {
+ "expected": [
+ "</option><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "option end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<option>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "option",
+ {}
+ ]
+ ],
+ "description": "option end-tag followed by option start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "option end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "option"
+ ]
+ ],
+ "description": "option end-tag at EOF"
+ },
+ {
+ "expected": [
+ "<colgroup><!--foo-->"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "colgroup start-tag followed by comment"
+ },
+ {
+ "expected": [
+ "<colgroup> foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "colgroup start-tag followed by space character"
+ },
+ {
+ "expected": [
+ "<colgroup>foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "colgroup start-tag followed by text"
+ },
+ {
+ "expected": [
+ "<colgroup><foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "colgroup start-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<table><col>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "table",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "col",
+ {}
+ ]
+ ],
+ "description": "first colgroup in a table with a col child"
+ },
+ {
+ "expected": [
+ "</colgroup><col>",
+ "<colgroup><col>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ],
+ [
+ "EmptyTag",
+ "http://www.w3.org/1999/xhtml",
+ "col",
+ {}
+ ]
+ ],
+ "description": "colgroup with a col child, following another colgroup"
+ },
+ {
+ "expected": [
+ "<colgroup></foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "colgroup start-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ "<colgroup>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup",
+ {}
+ ]
+ ],
+ "description": "colgroup start-tag at EOF"
+ },
+ {
+ "expected": [
+ "</colgroup><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "colgroup end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</colgroup> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "colgroup end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "colgroup end-tag followed by text"
+ },
+ {
+ "expected": [
+ "<foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "colgroup end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "colgroup end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "colgroup"
+ ]
+ ],
+ "description": "colgroup end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</thead><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "thead end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</thead> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "thead end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</thead>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "thead end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</thead><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "thead end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<tbody>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ]
+ ],
+ "description": "thead end-tag followed by tbody start-tag"
+ },
+ {
+ "expected": [
+ "<tfoot>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot",
+ {}
+ ]
+ ],
+ "description": "thead end-tag followed by tfoot start-tag"
+ },
+ {
+ "expected": [
+ "</thead></foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "thead end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ "</thead>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ]
+ ],
+ "description": "thead end-tag at EOF"
+ },
+ {
+ "expected": [
+ "<tbody><!--foo-->"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "tbody start-tag followed by comment"
+ },
+ {
+ "expected": [
+ "<tbody> foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "tbody start-tag followed by space character"
+ },
+ {
+ "expected": [
+ "<tbody>foo"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "tbody start-tag followed by text"
+ },
+ {
+ "expected": [
+ "<tbody><foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "tbody start-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<table><tr>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "table",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr",
+ {}
+ ]
+ ],
+ "description": "first tbody in a table with a tr child"
+ },
+ {
+ "expected": [
+ "<tbody><tr>",
+ "</tbody><tr>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr",
+ {}
+ ]
+ ],
+ "description": "tbody with a tr child, following another tbody"
+ },
+ {
+ "expected": [
+ "<tbody><tr>",
+ "</thead><tr>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "thead"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr",
+ {}
+ ]
+ ],
+ "description": "tbody with a tr child, following a thead"
+ },
+ {
+ "expected": [
+ "<tbody><tr>",
+ "</tfoot><tr>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr",
+ {}
+ ]
+ ],
+ "description": "tbody with a tr child, following a tfoot"
+ },
+ {
+ "expected": [
+ "<tbody></foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "tbody start-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ "<tbody>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ]
+ ],
+ "description": "tbody start-tag at EOF"
+ },
+ {
+ "expected": [
+ "</tbody><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "tbody end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</tbody> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "tbody end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</tbody>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "tbody end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</tbody><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "tbody end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<tbody>",
+ "</tbody>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ]
+ ],
+ "description": "tbody end-tag followed by tbody start-tag"
+ },
+ {
+ "expected": [
+ "<tfoot>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot",
+ {}
+ ]
+ ],
+ "description": "tbody end-tag followed by tfoot start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "tbody end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody"
+ ]
+ ],
+ "description": "tbody end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</tfoot><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "tfoot end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</tfoot> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "tfoot end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</tfoot>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "tfoot end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</tfoot><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "tfoot end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<tbody>",
+ "</tfoot>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tbody",
+ {}
+ ]
+ ],
+ "description": "tfoot end-tag followed by tbody start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "tfoot end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tfoot"
+ ]
+ ],
+ "description": "tfoot end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</tr><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "tr end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</tr> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "tr end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</tr>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "tr end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</tr><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "tr end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<tr>",
+ "</tr>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr",
+ {}
+ ]
+ ],
+ "description": "tr end-tag followed by tr start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "tr end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "tr"
+ ]
+ ],
+ "description": "tr end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</td><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "td end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</td> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "td end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</td>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "td end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</td><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "td end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<td>",
+ "</td>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "td",
+ {}
+ ]
+ ],
+ "description": "td end-tag followed by td start-tag"
+ },
+ {
+ "expected": [
+ "<th>",
+ "</td>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "th",
+ {}
+ ]
+ ],
+ "description": "td end-tag followed by th start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "td end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "td"
+ ]
+ ],
+ "description": "td end-tag at EOF"
+ },
+ {
+ "expected": [
+ "</th><!--foo-->"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ],
+ [
+ "Comment",
+ "foo"
+ ]
+ ],
+ "description": "th end-tag followed by comment"
+ },
+ {
+ "expected": [
+ "</th> foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ],
+ [
+ "Characters",
+ " foo"
+ ]
+ ],
+ "description": "th end-tag followed by space character"
+ },
+ {
+ "expected": [
+ "</th>foo"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ],
+ [
+ "Characters",
+ "foo"
+ ]
+ ],
+ "description": "th end-tag followed by text"
+ },
+ {
+ "expected": [
+ "</th><foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo",
+ {}
+ ]
+ ],
+ "description": "th end-tag followed by start-tag"
+ },
+ {
+ "expected": [
+ "<th>",
+ "</th>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "th",
+ {}
+ ]
+ ],
+ "description": "th end-tag followed by th start-tag"
+ },
+ {
+ "expected": [
+ "<td>",
+ "</th>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "td",
+ {}
+ ]
+ ],
+ "description": "th end-tag followed by td start-tag"
+ },
+ {
+ "expected": [
+ "</foo>"
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "foo"
+ ]
+ ],
+ "description": "th end-tag followed by end-tag"
+ },
+ {
+ "expected": [
+ ""
+ ],
+ "input": [
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "th"
+ ]
+ ],
+ "description": "th end-tag at EOF"
+ }
+ ]
+}
\ No newline at end of file
diff --git a/html5lib/tests/serializer-testdata/options.test b/html5lib/tests/serializer-testdata/options.test
new file mode 100644
index 00000000..eedcb3f0
--- /dev/null
+++ b/html5lib/tests/serializer-testdata/options.test
@@ -0,0 +1,265 @@
+{
+ "tests": [
+ {
+ "expected": [
+ "<span title='test 'with' quote_char'>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "test 'with' quote_char"
+ }
+ ]
+ ]
+ ],
+ "description": "quote_char=\"'\"",
+ "options": {
+ "quote_char": "'"
+ }
+ },
+ {
+ "expected": [
+ "<button disabled>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "button",
+ [
+ {
+ "namespace": null,
+ "name": "disabled",
+ "value": "disabled"
+ }
+ ]
+ ]
+ ],
+ "description": "quote_attr_values='always'",
+ "options": {
+ "quote_attr_values": "always"
+ }
+ },
+ {
+ "expected": [
+ "<div irrelevant>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ [
+ {
+ "namespace": null,
+ "name": "irrelevant",
+ "value": "irrelevant"
+ }
+ ]
+ ]
+ ],
+ "description": "quote_attr_values='always' with irrelevant",
+ "options": {
+ "quote_attr_values": "always"
+ }
+ },
+ {
+ "expected": [
+ "<div class=\"foo\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ [
+ {
+ "namespace": null,
+ "name": "class",
+ "value": "foo"
+ }
+ ]
+ ]
+ ],
+ "description": "non-minimized quote_attr_values='always'",
+ "options": {
+ "quote_attr_values": "always"
+ }
+ },
+ {
+ "expected": [
+ "<div class=foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ [
+ {
+ "namespace": null,
+ "name": "class",
+ "value": "foo"
+ }
+ ]
+ ]
+ ],
+ "description": "non-minimized quote_attr_values='legacy'",
+ "options": {
+ "quote_attr_values": "legacy"
+ }
+ },
+ {
+ "expected": [
+ "<div class=foo>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ [
+ {
+ "namespace": null,
+ "name": "class",
+ "value": "foo"
+ }
+ ]
+ ]
+ ],
+ "description": "non-minimized quote_attr_values='spec'",
+ "options": {
+ "quote_attr_values": "spec"
+ }
+ },
+ {
+ "expected": [
+ "<img />"
+ ],
+ "input": [
+ [
+ "EmptyTag",
+ "img",
+ {}
+ ]
+ ],
+ "description": "use_trailing_solidus=true with void element",
+ "options": {
+ "use_trailing_solidus": true
+ }
+ },
+ {
+ "expected": [
+ "<div>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ {}
+ ]
+ ],
+ "description": "use_trailing_solidus=true with non-void element",
+ "options": {
+ "use_trailing_solidus": true
+ }
+ },
+ {
+ "expected": [
+ "<div irrelevant=irrelevant>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ [
+ {
+ "namespace": null,
+ "name": "irrelevant",
+ "value": "irrelevant"
+ }
+ ]
+ ]
+ ],
+ "description": "minimize_boolean_attributes=false",
+ "options": {
+ "minimize_boolean_attributes": false
+ }
+ },
+ {
+ "expected": [
+ "<div irrelevant=\"\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "div",
+ [
+ {
+ "namespace": null,
+ "name": "irrelevant",
+ "value": ""
+ }
+ ]
+ ]
+ ],
+ "description": "minimize_boolean_attributes=false with empty value",
+ "options": {
+ "minimize_boolean_attributes": false
+ }
+ },
+ {
+ "expected": [
+ "<a title=\"a<b>c&d\">"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "a",
+ [
+ {
+ "namespace": null,
+ "name": "title",
+ "value": "a<b>c&d"
+ }
+ ]
+ ]
+ ],
+ "description": "escape less than signs in attribute values",
+ "options": {
+ "escape_lt_in_attrs": true
+ }
+ },
+ {
+ "expected": [
+ "<script>a<b>c&d"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "script",
+ {}
+ ],
+ [
+ "Characters",
+ "a<b>c&d"
+ ]
+ ],
+ "description": "rcdata",
+ "options": {
+ "escape_rcdata": true
+ }
+ }
+ ]
+}
\ No newline at end of file
diff --git a/html5lib/tests/serializer-testdata/whitespace.test b/html5lib/tests/serializer-testdata/whitespace.test
new file mode 100644
index 00000000..dac3a69e
--- /dev/null
+++ b/html5lib/tests/serializer-testdata/whitespace.test
@@ -0,0 +1,198 @@
+{
+ "tests": [
+ {
+ "expected": [
+ " foo"
+ ],
+ "input": [
+ [
+ "Characters",
+ "\t\r\n\f foo"
+ ]
+ ],
+ "description": "bare text with leading spaces",
+ "options": {
+ "strip_whitespace": true
+ }
+ },
+ {
+ "expected": [
+ "foo "
+ ],
+ "input": [
+ [
+ "Characters",
+ "foo \t\r\n\f"
+ ]
+ ],
+ "description": "bare text with trailing spaces",
+ "options": {
+ "strip_whitespace": true
+ }
+ },
+ {
+ "expected": [
+ "foo bar"
+ ],
+ "input": [
+ [
+ "Characters",
+ "foo \t\r\n\f bar"
+ ]
+ ],
+ "description": "bare text with inner spaces",
+ "options": {
+ "strip_whitespace": true
+ }
+ },
+ {
+ "expected": [
+ "<pre>\t\r\n\f foo \t\r\n\f bar \t\r\n\f</pre>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "pre",
+ {}
+ ],
+ [
+ "Characters",
+ "\t\r\n\f foo \t\r\n\f bar \t\r\n\f"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "pre"
+ ]
+ ],
+ "description": "text within <pre>",
+ "options": {
+ "strip_whitespace": true
+ }
+ },
+ {
+ "expected": [
+ "<pre>\t\r\n\f fo<span>o \t\r\n\f b</span>ar \t\r\n\f</pre>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "pre",
+ {}
+ ],
+ [
+ "Characters",
+ "\t\r\n\f fo"
+ ],
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "span",
+ {}
+ ],
+ [
+ "Characters",
+ "o \t\r\n\f b"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "span"
+ ],
+ [
+ "Characters",
+ "ar \t\r\n\f"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "pre"
+ ]
+ ],
+ "description": "text within <pre>, with inner markup",
+ "options": {
+ "strip_whitespace": true
+ }
+ },
+ {
+ "expected": [
+ "<textarea>\t\r\n\f foo \t\r\n\f bar \t\r\n\f</textarea>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "textarea",
+ {}
+ ],
+ [
+ "Characters",
+ "\t\r\n\f foo \t\r\n\f bar \t\r\n\f"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "textarea"
+ ]
+ ],
+ "description": "text within <textarea>",
+ "options": {
+ "strip_whitespace": true
+ }
+ },
+ {
+ "expected": [
+ "<script>\t\r\n\f foo \t\r\n\f bar \t\r\n\f</script>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "script",
+ {}
+ ],
+ [
+ "Characters",
+ "\t\r\n\f foo \t\r\n\f bar \t\r\n\f"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "script"
+ ]
+ ],
+ "description": "text within <script>",
+ "options": {
+ "strip_whitespace": true
+ }
+ },
+ {
+ "expected": [
+ "<style>\t\r\n\f foo \t\r\n\f bar \t\r\n\f</style>"
+ ],
+ "input": [
+ [
+ "StartTag",
+ "http://www.w3.org/1999/xhtml",
+ "style",
+ {}
+ ],
+ [
+ "Characters",
+ "\t\r\n\f foo \t\r\n\f bar \t\r\n\f"
+ ],
+ [
+ "EndTag",
+ "http://www.w3.org/1999/xhtml",
+ "style"
+ ]
+ ],
+ "description": "text within <style>",
+ "options": {
+ "strip_whitespace": true
+ }
+ }
+ ]
+}
\ No newline at end of file
diff --git a/html5lib/tests/support.py b/html5lib/tests/support.py
index 54a64a85..6e6a916b 100644
--- a/html5lib/tests/support.py
+++ b/html5lib/tests/support.py
@@ -69,8 +69,8 @@
}
-def get_data_files(subdirectory, files='*.dat'):
- return sorted(glob.glob(os.path.join(test_dir, subdirectory, files)))
+def get_data_files(subdirectory, files='*.dat', search_dir=test_dir):
+ return sorted(glob.glob(os.path.join(search_dir, subdirectory, files)))
class DefaultDict(dict):
diff --git a/html5lib/tests/test_serializer.py b/html5lib/tests/test_serializer.py
index af76075e..93276267 100644
--- a/html5lib/tests/test_serializer.py
+++ b/html5lib/tests/test_serializer.py
@@ -1,17 +1,14 @@
from __future__ import absolute_import, division, unicode_literals
+import os
import json
-import unittest
-from .support import get_data_files
+import pytest
-try:
- unittest.TestCase.assertEqual
-except AttributeError:
- unittest.TestCase.assertEqual = unittest.TestCase.assertEquals
+from .support import get_data_files
-import html5lib
from html5lib import constants
+from html5lib.filters.lint import Filter as Lint
from html5lib.serializer import HTMLSerializer, serialize
from html5lib.treewalkers._base import TreeWalker
@@ -82,7 +79,7 @@ def _convertAttrib(self, attribs):
def serialize_html(input, options):
options = dict([(str(k), v) for k, v in options.items()])
- stream = JsonWalker(input)
+ stream = Lint(JsonWalker(input), False)
serializer = HTMLSerializer(alphabetical_attributes=True, **options)
return serializer.render(stream, options.get("encoding", None))
@@ -100,77 +97,118 @@ def runSerializerTest(input, expected, options):
assert False, "Expected: %s, Received: %s" % (expected, result)
-class EncodingTestCase(unittest.TestCase):
- def throwsWithLatin1(self, input):
- self.assertRaises(UnicodeEncodeError, serialize_html, input, {"encoding": "iso-8859-1"})
+def throwsWithLatin1(input):
+ with pytest.raises(UnicodeEncodeError):
+ serialize_html(input, {"encoding": "iso-8859-1"})
+
+
+def testDoctypeName():
+ throwsWithLatin1([["Doctype", "\u0101"]])
+
+
+def testDoctypePublicId():
+ throwsWithLatin1([["Doctype", "potato", "\u0101"]])
+
+
+def testDoctypeSystemId():
+ throwsWithLatin1([["Doctype", "potato", "potato", "\u0101"]])
+
+
+def testCdataCharacters():
+ runSerializerTest([["StartTag", "http://www.w3.org/1999/xhtml", "style", {}], ["Characters", "\u0101"]],
+ ["<style>ā"], {"encoding": "iso-8859-1"})
+
+
+def testCharacters():
+ runSerializerTest([["Characters", "\u0101"]],
+ ["ā"], {"encoding": "iso-8859-1"})
+
+
+def testStartTagName():
+ throwsWithLatin1([["StartTag", "http://www.w3.org/1999/xhtml", "\u0101", []]])
+
+
+def testAttributeName():
+ throwsWithLatin1([["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": None, "name": "\u0101", "value": "potato"}]]])
+
+
+def testAttributeValue():
+ runSerializerTest([["StartTag", "http://www.w3.org/1999/xhtml", "span",
+ [{"namespace": None, "name": "potato", "value": "\u0101"}]]],
+ ["<span potato=ā>"], {"encoding": "iso-8859-1"})
- def testDoctypeName(self):
- self.throwsWithLatin1([["Doctype", "\u0101"]])
- def testDoctypePublicId(self):
- self.throwsWithLatin1([["Doctype", "potato", "\u0101"]])
+def testEndTagName():
+ throwsWithLatin1([["EndTag", "http://www.w3.org/1999/xhtml", "\u0101"]])
- def testDoctypeSystemId(self):
- self.throwsWithLatin1([["Doctype", "potato", "potato", "\u0101"]])
- def testCdataCharacters(self):
- runSerializerTest([["StartTag", "http://www.w3.org/1999/xhtml", "style", {}], ["Characters", "\u0101"]],
- ["<style>ā"], {"encoding": "iso-8859-1"})
+def testComment():
+ throwsWithLatin1([["Comment", "\u0101"]])
- def testCharacters(self):
- runSerializerTest([["Characters", "\u0101"]],
- ["ā"], {"encoding": "iso-8859-1"})
- def testStartTagName(self):
- self.throwsWithLatin1([["StartTag", "http://www.w3.org/1999/xhtml", "\u0101", []]])
+@pytest.mark.parametrize("c", list("\t\n\u000C\x20\r\"'=<>`"))
+def testSpecQuoteAttribute(c):
+ input_ = [["StartTag", "http://www.w3.org/1999/xhtml", "span",
+ [{"namespace": None, "name": "foo", "value": c}]]]
+ if c == '"':
+ output_ = ["<span foo='%s'>" % c]
+ else:
+ output_ = ['<span foo="%s">' % c]
+ options_ = {"quote_attr_values": "spec"}
+ runSerializerTest(input_, output_, options_)
- def testEmptyTagName(self):
- self.throwsWithLatin1([["EmptyTag", "http://www.w3.org/1999/xhtml", "\u0101", []]])
- def testAttributeName(self):
- self.throwsWithLatin1([["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": None, "name": "\u0101", "value": "potato"}]]])
+@pytest.mark.parametrize("c", list("\t\n\u000C\x20\r\"'=<>`"
+ "\x00\x01\x02\x03\x04\x05\x06\x07\x08\t\n"
+ "\x0b\x0c\r\x0e\x0f\x10\x11\x12\x13\x14\x15"
+ "\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f"
+ "\x20\x2f\x60\xa0\u1680\u180e\u180f\u2000"
+ "\u2001\u2002\u2003\u2004\u2005\u2006\u2007"
+ "\u2008\u2009\u200a\u2028\u2029\u202f\u205f"
+ "\u3000"))
+def testLegacyQuoteAttribute(c):
+ input_ = [["StartTag", "http://www.w3.org/1999/xhtml", "span",
+ [{"namespace": None, "name": "foo", "value": c}]]]
+ if c == '"':
+ output_ = ["<span foo='%s'>" % c]
+ else:
+ output_ = ['<span foo="%s">' % c]
+ options_ = {"quote_attr_values": "legacy"}
+ runSerializerTest(input_, output_, options_)
- def testAttributeValue(self):
- runSerializerTest([["StartTag", "http://www.w3.org/1999/xhtml", "span",
- [{"namespace": None, "name": "potato", "value": "\u0101"}]]],
- ["<span potato=ā>"], {"encoding": "iso-8859-1"})
- def testEndTagName(self):
- self.throwsWithLatin1([["EndTag", "http://www.w3.org/1999/xhtml", "\u0101"]])
+@pytest.fixture
+def lxml_parser():
+ return etree.XMLParser(resolve_entities=False)
- def testComment(self):
- self.throwsWithLatin1([["Comment", "\u0101"]])
+@pytest.mark.skipif("lxml" not in optionals_loaded, reason="lxml not importable")
+def testEntityReplacement(lxml_parser):
+ doc = '<!DOCTYPE html SYSTEM "about:legacy-compat"><html>β</html>'
+ tree = etree.fromstring(doc, parser=lxml_parser).getroottree()
+ result = serialize(tree, tree="lxml", omit_optional_tags=False)
+ assert result == '<!DOCTYPE html SYSTEM "about:legacy-compat"><html>\u03B2</html>'
-if "lxml" in optionals_loaded:
- class LxmlTestCase(unittest.TestCase):
- def setUp(self):
- self.parser = etree.XMLParser(resolve_entities=False)
- self.treewalker = html5lib.getTreeWalker("lxml")
- self.serializer = HTMLSerializer()
- def testEntityReplacement(self):
- doc = """<!DOCTYPE html SYSTEM "about:legacy-compat"><html>β</html>"""
- tree = etree.fromstring(doc, parser=self.parser).getroottree()
- result = serialize(tree, tree="lxml", omit_optional_tags=False)
- self.assertEqual("""<!DOCTYPE html SYSTEM "about:legacy-compat"><html>\u03B2</html>""", result)
+@pytest.mark.skipif("lxml" not in optionals_loaded, reason="lxml not importable")
+def testEntityXML(lxml_parser):
+ doc = '<!DOCTYPE html SYSTEM "about:legacy-compat"><html>></html>'
+ tree = etree.fromstring(doc, parser=lxml_parser).getroottree()
+ result = serialize(tree, tree="lxml", omit_optional_tags=False)
+ assert result == '<!DOCTYPE html SYSTEM "about:legacy-compat"><html>></html>'
- def testEntityXML(self):
- doc = """<!DOCTYPE html SYSTEM "about:legacy-compat"><html>></html>"""
- tree = etree.fromstring(doc, parser=self.parser).getroottree()
- result = serialize(tree, tree="lxml", omit_optional_tags=False)
- self.assertEqual("""<!DOCTYPE html SYSTEM "about:legacy-compat"><html>></html>""", result)
- def testEntityNoResolve(self):
- doc = """<!DOCTYPE html SYSTEM "about:legacy-compat"><html>β</html>"""
- tree = etree.fromstring(doc, parser=self.parser).getroottree()
- result = serialize(tree, tree="lxml", omit_optional_tags=False,
- resolve_entities=False)
- self.assertEqual("""<!DOCTYPE html SYSTEM "about:legacy-compat"><html>β</html>""", result)
+@pytest.mark.skipif("lxml" not in optionals_loaded, reason="lxml not importable")
+def testEntityNoResolve(lxml_parser):
+ doc = '<!DOCTYPE html SYSTEM "about:legacy-compat"><html>β</html>'
+ tree = etree.fromstring(doc, parser=lxml_parser).getroottree()
+ result = serialize(tree, tree="lxml", omit_optional_tags=False,
+ resolve_entities=False)
+ assert result == '<!DOCTYPE html SYSTEM "about:legacy-compat"><html>β</html>'
def test_serializer():
- for filename in get_data_files('serializer', '*.test'):
+ for filename in get_data_files('serializer-testdata', '*.test', os.path.dirname(__file__)):
with open(filename) as fp:
tests = json.load(fp)
for index, test in enumerate(tests['tests']):