def test_string_contains_lone_surrogates(self):
self.assertFalse(
StringEncoding.string_contains_lone_surrogates(u"abc"))
self.assertFalse(
StringEncoding.string_contains_lone_surrogates(u"\uABCD"))
self.assertFalse(
StringEncoding.string_contains_lone_surrogates(u"\N{SNOWMAN}"))
# This behaves differently in Py2 when freshly parsed and read from a .pyc file,
# but it seems to be a marshalling bug in Py2, which doesn't hurt us in Cython.
if sys.version_info[0] != 2:
self.assertTrue(
StringEncoding.string_contains_lone_surrogates(
u"\uD800\uDFFF"))
# In Py2 with 16bit Unicode, the following is indistinguishable from the 32bit character.
obfuscated_surrogate_pair = (u"\uDFFF" + "\uD800")[::-1]
if sys.version_info[0] == 2 and sys.maxunicode == 65565:
self.assertFalse(
StringEncoding.string_contains_lone_surrogates(
obfuscated_surrogate_pair))
else:
self.assertTrue(
StringEncoding.string_contains_lone_surrogates(
obfuscated_surrogate_pair))
self.assertTrue(
StringEncoding.string_contains_lone_surrogates(u"\uD800"))
self.assertTrue(
StringEncoding.string_contains_lone_surrogates(u"\uDFFF"))
self.assertTrue(
StringEncoding.string_contains_lone_surrogates(u"\uDFFF\uD800"))
self.assertTrue(
StringEncoding.string_contains_lone_surrogates(u"\uD800x\uDFFF"))
def synthesize_defnodes(self):
"""
Create the __signatures__ dict of PyCFunctionNode specializations.
"""
if isinstance(self.nodes[0], CFuncDefNode):
nodes = [node.py_func for node in self.nodes]
else:
nodes = self.nodes
signatures = [
StringEncoding.EncodedString(node.specialized_signature_string)
for node in nodes
]
keys = [
ExprNodes.StringNode(node.pos, value=sig)
for node, sig in zip(nodes, signatures)
]
values = [
ExprNodes.PyCFunctionNode.from_defnode(node, True)
for node in nodes
]
self.__signatures__ = ExprNodes.DictNode.from_pairs(
self.pos, zip(keys, values))
self.specialized_pycfuncs = values
for pycfuncnode in values:
pycfuncnode.is_specialization = True
def visit_Constant(self, node):
if isinstance(node.value, int):
return IntNode(self.loc(node), value=str(node.value))
if isinstance(node.value, float):
# We won't have c float, we map floating types to double
return FloatNode(self.loc(node), value=str(node.value))
if isinstance(node.value, str):
if node.kind == "u":
return UnicodeNode(self.loc(node), value=node.value, bytes_value=None)
else:
return StringNode(
self.loc(node),
value=node.value,
unicode_value=StringEncoding.EncodedString(node.value),
)
if (
isinstance(Ellipsis, type)
and isinstance(node.value, Ellipsis)
or isinstance(node.value, type(Ellipsis))
):
return EllipsisNode(self.loc(node))
if isinstance(node.value, bytes):
return BytesNode(self.loc(node), value=node.s)
if node.value is None:
return NoneNode(self.loc(node))
raise NotImplementedError("Unknown constant value: %s" % node)
def test_string_contains_surrogates(self):
self.assertFalse(StringEncoding.string_contains_surrogates(u"abc"))
self.assertFalse(StringEncoding.string_contains_surrogates(u"\uABCD"))
self.assertFalse(
StringEncoding.string_contains_surrogates(u"\N{SNOWMAN}"))
self.assertTrue(StringEncoding.string_contains_surrogates(u"\uD800"))
self.assertTrue(StringEncoding.string_contains_surrogates(u"\uDFFF"))
self.assertTrue(
StringEncoding.string_contains_surrogates(u"\uD800\uDFFF"))
self.assertTrue(
StringEncoding.string_contains_surrogates(u"\uDFFF\uD800"))
self.assertTrue(
StringEncoding.string_contains_surrogates(u"\uD800x\uDFFF"))
def visit_CallNode(self, node):
if isinstance(node.function, ExprNodes.NameNode) and \
node.function.name == u"DAAPObject":
# Make sure we only convert DAAPObject(x, y) calls, nothing more.
if len(node.args) == 2:
code = daap_data.dmap_names[node.args[0].value]
itype = daap_data.dmap_code_types[code][1]
node.function.name = self.context.intern_ustring(
u"SpeedyDAAPObject")
node.args[0] = ExprNodes.StringNode(
node.pos, value=StringEncoding.BytesLiteral(code))
node.args.insert(1, ExprNodes.IntNode(
node.pos, value=str(itype)))
# Visit method body.
self.visitchildren(node)
return node
def visit_Str(self, node):
return StringNode(
self.loc(node),
value=node.s,
unicode_value=StringEncoding.EncodedString(node.s),
)