def test_decode_utf8_1():
"""
Counterpart of test_encode_utf8_1.
Tests decoding of some special cases:
1 - c080 must be decoded as byte 00
2 - eda081edb080 must be decoded as \uD801\uDC00 (representing U+10400)
"""
str1 = '31c08032'
str2 = decode_modified_utf8(binascii.unhexlify(str1))
assert u'1\x002' == str2
str1 = 'eda081edb080'
str2 = decode_modified_utf8(binascii.unhexlify(str1))
assert u'\uD801\uDC00' == str2
def _from_io(self, fio):
# Reads in the ConstantPool (constant_pool in the JVM Spec)
constant_pool_count = unpack('>H', fio.read(2))[0]
# Pull this locally so CPython doesn't do a lookup each time.
read = fio.read
while constant_pool_count > 1:
constant_pool_count -= 1
# The 1-byte prefix identifies the type of constant.
tag = unpack('>B', read(1))[0]
if tag == 1:
# CONSTANT_Utf8_info, a length prefixed UTF-8-ish string.
length = unpack('>H', read(2))[0]
self.append((tag, decode_modified_utf8(read(length))))
else:
# Every other constant type is trivial.
fmt, size = _constant_fmts[tag]
self.append((tag,) + unpack(fmt, read(size)))
if tag in (5, 6):
# LONG (5) and DOUBLE (6) count as two entries in the
# pool.
self.append(None)
constant_pool_count -= 1
def unpack(self, fio):
"""
Read the ConstantPool from the file-like object `fio`.
.. note::
Advanced usage only. You will typically never need to call this
method as it will be called for you when loading a ClassFile.
:param fio: Any file-like object providing `read()`
"""
# Reads in the ConstantPool (constant_pool in the JVM Spec)
constant_pool_count = unpack('>H', fio.read(2))[0]
# Pull this locally so CPython doesn't do a lookup each time.
read = fio.read
while constant_pool_count > 1:
constant_pool_count -= 1
# The 1-byte prefix identifies the type of constant.
tag = unpack('>B', read(1))[0]
if tag == 1:
# CONSTANT_Utf8_info, a length prefixed UTF-8-ish string.
length = unpack('>H', read(2))[0]
self.append((tag, decode_modified_utf8(read(length))))
else:
# Every other constant type is trivial.
fmt, size = _constant_fmts[tag]
self.append((tag,) + unpack(fmt, read(size)))
if tag in (5, 6):
# LONG (5) and DOUBLE (6) count as two entries in the
# pool.
self.append(None)
constant_pool_count -= 1
def test_decode_utf8_1():
"""
Counterpart of test_encode_utf8_1.
Tests decoding of some special cases:
1 - c080 must be decoded as byte 00
2 - eda081edb080 must be decoded as \uD801\uDC00 (representing U+10400)
"""
pairs = ((b'\x31\xc0\x80\x32', '1\x002'), (b'\xed\xa0\x81\xed\xb0\x80',
'\uD801\uDC00'))
for original, decoded in pairs:
assert decode_modified_utf8(original) == decoded
def test_decode_utf8_1():
"""
Counterpart of test_encode_utf8_1.
Tests decoding of some special cases:
1 - c080 must be decoded as byte 00
2 - eda081edb080 must be decoded as \uD801\uDC00 (representing U+10400)
"""
pairs = (
(b'\x31\xc0\x80\x32', '1\x002'),
(b'\xed\xa0\x81\xed\xb0\x80', '\uD801\uDC00')
)
for original, decoded in pairs:
assert decode_modified_utf8(original) == decoded
def unpack(self, fio):
"""
Read the ConstantPool from the file-like object `fio`.
.. note::
Advanced usage only. You will typically never need to call this
method as it will be called for you when loading a ClassFile.
:param fio: Any file-like object providing `read()`
"""
# Reads in the ConstantPool (constant_pool in the JVM Spec)
constant_pool_count = unpack('>H', fio.read(2))[0]
# Pull this locally so CPython doesn't do a lookup each time.
read = fio.read
while constant_pool_count > 1:
constant_pool_count -= 1
# The 1-byte prefix identifies the type of constant.
tag = ord(read(1))
if tag == 1:
# CONSTANT_Utf8_info, a length prefixed UTF-8-ish string.
# Only attempt to properly decode the MUTF8 if it fails
# regular UTF8 decoding, which overs huge time savings over
# large JARs.
utf8_str = read(unpack('>H', read(2))[0])
try:
utf8_str = utf8_str.decode('utf8')
except UnicodeDecodeError:
utf8_str = decode_modified_utf8(utf8_str)
self.append((tag, utf8_str))
else:
# Every other constant type is trivial.
fmt, size = _constant_fmts[tag]
self.append((tag, *unpack(fmt, read(size))))
if tag == 5 or tag == 6:
# LONG (5) and DOUBLE (6) count as two entries in the
# pool.
self.append(None)
constant_pool_count -= 1