def test_strip():
xml = """<html>
<head><title>The Title</title></head>
<body>
<h1>Header 1</h1>
<div class="1">
<h2>Header 1.1</h2>
<h2>Header 1.2</h2>
<div class="2">
<h3>Header 1.2.1</h3>
<h3>Header 1.2.2</h3>
<div class="2">
<p>Hello, World!</p>
</div>
<h3>Header 1.2.3</h3>
</div>
<h2>Header 1.3</h2>
</div>
</body>
</html>"""
xml_stream = XML(xml, fragment=False)
assert type(xml_stream) is ElementStream
xml_stream_stripped = xml_stream.strip(levels=6)
assert type(xml_stream_stripped) is ElementStream
from kid import XMLSerializer
serializer = XMLSerializer(decl=False)
xml_stripped = serializer.serialize(xml_stream_stripped, fragment=True)
assert xml_stripped == 'Hello, World!'
def test_strip():
xml = """<html>
<head><title>The Title</title></head>
<body>
<h1>Header 1</h1>
<div class="1">
<h2>Header 1.1</h2>
<h2>Header 1.2</h2>
<div class="2">
<h3>Header 1.2.1</h3>
<h3>Header 1.2.2</h3>
<div class="2">
<p>Hello, World!</p>
</div>
<h3>Header 1.2.3</h3>
</div>
<h2>Header 1.3</h2>
</div>
</body>
</html>"""
xml_stream = XML(xml, fragment=False)
assert type(xml_stream) is ElementStream
xml_stream_stripped = xml_stream.strip(levels=6)
assert type(xml_stream_stripped) is ElementStream
from kid import XMLSerializer
serializer = XMLSerializer(decl=False)
xml_stripped = serializer.serialize(xml_stream_stripped, fragment=True)
assert xml_stripped == 'Hello, World!'
def test_expand():
doc = XML("<doc><hello>world</hello></doc>", fragment=False)
assert type(doc) is ElementStream
doc = doc.expand()
assert type(doc) == type(Element('doc'))
assert doc.tag == 'doc'
assert doc[0].tag == 'hello'
assert doc[0].text == 'world'
def test_xml_with_entity_map():
xml = ' '
assert list(XML(xml))[0][1] == u'\xa0'
xml = '&codswallop;'
try:
e = list(XML(xml))
except Exception, e:
e = str(e)
def test_expand():
doc = XML("<doc><hello>world</hello></doc>", fragment=False)
assert type(doc) is ElementStream
doc = doc.expand()
assert type(doc) == type(Element('doc'))
assert doc.tag == 'doc'
assert doc[0].tag == 'hello'
assert doc[0].text == 'world'
def transform(self, stream=None, filters=[]):
"""
Execute the template and apply any match transformations.
If stream is specified, it must be one of the following:
Element
A kid.Element.
ElementStream
An `pull.ElementStream` instance or other iterator that yields
stream events.
string
A file or URL unless the string starts with
'<' in which case it is considered an XML document
and processed as if it had been an Element.
By default, the `pull` method is called to obtain the stream.
"""
if stream is None:
stream = self.pull()
elif isinstance(stream, basestring):
if xml_sniff(stream):
stream = XML(stream, fragment=False)
else:
stream = document(stream)
elif hasattr(stream, 'tag'):
stream = ElementStream(stream)
else:
stream = ElementStream.ensure(stream)
for f in filters + self._filters:
stream = f(stream, self)
return stream
def test_xml_type():
doc = XML("<doc>hello world</doc>", fragment=False)
assert type(doc) is ElementStream
doc = XML("hello world", fragment=True)
assert type(doc) is ElementStream
xml_stripped = serializer.serialize(xml_stream_stripped, fragment=True)
assert xml_stripped == 'Hello, World!'
def test_xml_with_entity_map():
xml = ' '
assert list(XML(xml))[0][1] == u'\xa0'
xml = '&codswallop;'
try:
e = list(XML(xml))
except Exception, e:
e = str(e)
assert 'undefined entity &codswallop;' in e
xml = ' , &codswallop;!'
entity_map = {'nbsp': u'Hello', 'codswallop': u'World'}
assert list(XML(xml, entity_map=entity_map))[0][1] == u'Hello, World!'
def test_load_template_with_entity_map():
xml = '<html> </html>'
s = load_template(xml).serialize(encoding='ascii')
assert s.endswith('<html> </html>')
xml = '<html>&codswallop;</html>'
try:
e = load_template(xml).serialize(encoding='ascii')
except Exception, e:
e = str(e)
assert 'undefined entity &codswallop;' in e
xml = '<html> , &codswallop;!</html>'
entity_map = {'nbsp': u'Hello', 'codswallop': u'World'}
s = load_template(xml, entity_map=entity_map).serialize(encoding='ascii')
