def test_external_parameter_entity_definition(self):
with self.subTest("SYSTEM"):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root [<!ENTITY % entity SYSTEM 'uri'>]>")
self.assertEqual("uri",
document.parameter_entities["entity"].system_URI)
self.assertEqual(Entity.Type.PARAMETER,
document.parameter_entities["entity"].type)
self.assertEqual(True,
document.parameter_entities["entity"].external)
self.assertEqual(True,
document.parameter_entities["entity"].parsed)
with self.subTest("PUBLIC"):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root [<!ENTITY % entity PUBLIC 'uri1' 'uri2'>]>")
self.assertEqual("uri1",
document.parameter_entities["entity"].public_URI)
self.assertEqual("uri2",
document.parameter_entities["entity"].system_URI)
self.assertEqual(Entity.Type.PARAMETER,
document.parameter_entities["entity"].type)
self.assertEqual(True,
document.parameter_entities["entity"].external)
self.assertEqual(True,
document.parameter_entities["entity"].parsed)
def test_no_text_in_internal_subset(self):
document = Document("")
with self.assertRaises(XMLError):
document._Document__parse_doctype_declaration("""<!DOCTYPE root [
<!ENTITY % entity 'blah, blahblah'>'
some disallowed text
]>""")
def test_unparsed_entity_definition(self):
with self.subTest("SYSTEM"):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root [<!ENTITY entity SYSTEM 'uri' NDATA notation>]>"
)
self.assertEqual("uri",
document.general_entities["entity"].system_URI)
self.assertEqual(Entity.Type.GENERAL,
document.general_entities["entity"].type)
self.assertEqual(True,
document.general_entities["entity"].external)
self.assertEqual(False, document.general_entities["entity"].parsed)
self.assertEqual("notation",
document.general_entities["entity"].notation)
with self.subTest("PUBLIC"):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root [<!ENTITY entity PUBLIC 'uri1' 'uri2' NDATA notation>]>"
)
self.assertEqual("uri1",
document.general_entities["entity"].public_URI)
self.assertEqual("uri2",
document.general_entities["entity"].system_URI)
self.assertEqual(Entity.Type.GENERAL,
document.general_entities["entity"].type)
self.assertEqual(True,
document.general_entities["entity"].external)
self.assertEqual(False, document.general_entities["entity"].parsed)
self.assertEqual("notation",
document.general_entities["entity"].notation)
def test_forbidden_name_characters(self):
for char in ["\u00d7", "\u00f7", "\u037e", "\u2030"]:
with self.subTest(char):
document = Document("")
with self.assertRaises(XMLError):
document._Document__parse_doctype_declaration(
f"<!DOCTYPE root{char}>")
def test_forbidden_name_start_characters(self):
for char in ["-", ".", "0", "7", "\u0305"]:
with self.subTest(char):
document = Document("")
with self.assertRaises(XMLError):
document._Document__parse_doctype_declaration(
f"<!DOCTYPE {char}root>")
def test_internal_parameter_entity_definition(self):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root [<!ENTITY % entity 'blah'>]>")
self.assertEqual("blah",
document.parameter_entities["entity"].expansion_text)
self.assertEqual(Entity.Type.PARAMETER,
document.parameter_entities["entity"].type)
self.assertEqual(False, document.parameter_entities["entity"].external)
def test_root_name_parsing(self):
with self.subTest("With no subset"):
document = Document("")
document._Document__parse_doctype_declaration("<!DOCTYPE root>")
self.assertEqual("root", document.dtd_name)
with self.subTest("With empty internal subset"):
document = Document("")
document._Document__parse_doctype_declaration("<!DOCTYPE root []>")
self.assertEqual("root", document.dtd_name)
def test_parameter_reference_within_markup(self):
with self.subTest("ELEMENT declaration"):
document = Document("")
with self.assertRaises(XMLError):
document._Document__parse_doctype_declaration(
"""<!DOCTYPE root [
<!ENTITY % entity 'blah, blahblah'>'
<!ELEMENT root (%entity;)>
]>""")
with self.subTest("NOTATION declaration"):
document = Document("")
with self.assertRaises(XMLError):
document._Document__parse_doctype_declaration(
"""<!DOCTYPE root [
<!ENTITY % entity 'SYSTEM \"blahblah\"'>'
<!NOTATION blah %entity;>
]>""")
with self.subTest("ATTLIST declaration"):
document = Document("")
with self.assertRaises(XMLError):
document._Document__parse_doctype_declaration(
"""<!DOCTYPE root [
<!ENTITY % entity 'CDATA #REQUIRED'>'
<!ATTLIST root %entity;>
]>""")
def test_name_collisions(self):
with self.subTest(
"General and parameter entities occupy different namespaces"):
document = Document("")
document._Document__parse_doctype_declaration("""<!DOCTYPE root [
<!ENTITY % entity 'SOME TEXT'>
<!ENTITY entity 'SOME OTHER TEXT'>
]>""")
self.assertEqual(
"SOME TEXT",
document.parameter_entities["entity"].expansion_text)
self.assertEqual(
"SOME OTHER TEXT",
document.general_entities["entity"].expansion_text)
with self.subTest("General entities"):
document = Document("")
document._Document__parse_doctype_declaration("""<!DOCTYPE root [
<!ENTITY entity 'SOME TEXT'>
<!ENTITY entity 'SOME OTHER TEXT'>
]>""")
self.assertEqual(
"SOME TEXT",
document.general_entities["entity"].expansion_text)
with self.subTest("Parameter entities"):
document = Document("")
document._Document__parse_doctype_declaration("""<!DOCTYPE root [
<!ENTITY % entity 'SOME TEXT'>
<!ENTITY % entity 'SOME OTHER TEXT'>
]>""")
self.assertEqual(
"SOME TEXT",
document.parameter_entities["entity"].expansion_text)
def test_parameter_entity_expansion(self):
with self.subTest("In DTD"):
document = Document("")
document._Document__parse_doctype_declaration("""<!DOCTYPE root [
<!ENTITY % entity '<!ENTITY entity2 "SOME TEXT">'>
%entity;
]>""")
self.assertEqual(
"SOME TEXT",
document.general_entities["entity2"].expansion_text)
with self.subTest("In general entity"):
document = Document("")
document._Document__parse_doctype_declaration("""<!DOCTYPE root [
<!ENTITY % entity 'SOME TEXT'>
<!ENTITY entity2 'SOME %entity; OTHER TEXT'>
]>""")
self.assertEqual(
"SOME TEXT",
document.parameter_entities["entity"].expansion_text)
self.assertEqual(
"SOME SOME TEXT OTHER TEXT",
document.general_entities["entity2"].expansion_text)
def test_external_declaration_parsing(self):
with self.subTest("SYSTEM"):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root SYSTEM 'uri'>")
self.assertEqual("root", document.dtd_name)
self.assertEqual("uri", document.external_system_uri)
with self.subTest("PUBLIC"):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root PUBLIC 'uri1' 'uri2'>")
self.assertEqual("root", document.dtd_name)
self.assertEqual("uri1", document.external_public_uri)
self.assertEqual("uri2", document.external_system_uri)
with self.subTest("SYSTEM with internal subset"):
document = Document("")
document._Document__parse_doctype_declaration(
"<!DOCTYPE root SYSTEM 'uri' []>")
self.assertEqual("root", document.dtd_name)
self.assertEqual("uri", document.external_system_uri)