def parse(xml: str) -> Document:
# Normalise whitespace
xml = xml.replace("\u000d\u000a", "\u000a")
xml = xml.replace("\u000d", "\u000a")
# Parse document
document = Document(xml)
document.parse()
return document
def test_with_xmldeclaration(self):
document = Document(
"<?xml version='1.0' encoding='utf-8' standalone='yes'?> <root></root>"
)
document.parse()
self.assertEqual("1.0", document.version)
self.assertEqual("utf-8", document.encoding)
self.assertEqual(True, document.standalone)
self.assertIsInstance(document.root, Element)
self.assertEqual(document.root.name, "root")
def test_without_xmldeclaration(self):
with self.subTest("Open tag"):
document = Document("<root></root>")
document.parse()
self.assertIsInstance(document.root, Element)
self.assertEqual(document.root.name, "root")
with self.subTest("Closed tag"):
document = Document("<root/>")
document.parse()
self.assertIsInstance(document.root, Element)
self.assertEqual(document.root.name, "root")
def test_with_empty_dtd(self):
document = Document("""
<?xml version='1.0' encoding='utf-8' standalone='yes'?>
<!DOCTYPE root []>
<root></root>
""")
document.parse()
self.assertEqual("1.0", document.version)
self.assertEqual("utf-8", document.encoding)
self.assertEqual(True, document.standalone)
self.assertIsInstance(document.root, Element)
self.assertEqual(document.root.name, "root")
def test_with_fluff(self):
document = Document("""
<?xml version='1.0' encoding='utf-8' standalone='yes'?>
<!DOCTYPE root []>
<?Target A rogue processing instruction?>
<!-- And even a comment! -->
<root></root>
<!-- Another comment -->
<?Target And more PIs?>
<?Target?>
""")
document.parse()
self.assertEqual("1.0", document.version)
self.assertEqual("utf-8", document.encoding)
self.assertEqual(True, document.standalone)
self.assertIsInstance(document.root, Element)
self.assertEqual(document.root.name, "root")
self.assertEqual(3, len(document.processing_instructions))
def test_no_superfluous_characters(self):
with self.subTest("Before xml declaration"):
document = Document("some text<?xml version='1.0'?><root></root>")
with self.assertRaises(XMLError):
document.parse()
with self.subTest("Between xml declaration and dtd"):
document = Document(
"<?xml version='1.0'?>some text<!DOCTYPE root []><root></root>"
)
with self.assertRaises(XMLError):
document.parse()
with self.subTest("Between dtd and root element"):
document = Document(
"<?xml version='1.0'?><!DOCTYPE root []>some text<root></root>"
)
with self.assertRaises(XMLError):
document.parse()
with self.subTest("After root element"):
document = Document(
"<?xml version='1.0'?><!DOCTYPE root []><root></root>some text"
)
with self.assertRaises(XMLError):
document.parse()
def test_only_one_root_element(self):
document = Document(
"<?xml version='1.0' encoding='utf-8' standalone='yes'?> <root1></root1> <root2></root2>"
)
with self.assertRaises(XMLError):
document.parse()
