def get_encoder(self):
"""
Returns a HPACK encoder set up for responses.
"""
e = Encoder()
e.huffman_coder = HuffmanEncoder(REQUEST_CODES, REQUEST_CODES_LENGTH)
return e
def build_headers_frame(headers, encoder=None):
f = HeadersFrame(1)
e = encoder
if e is None:
e = Encoder()
e.huffman_coder = HuffmanEncoder(REQUEST_CODES, REQUEST_CODES_LENGTH)
f.data = e.encode(headers)
f.flags.add('END_HEADERS')
return f
def test_request_huffman_encode(self):
encoder = HuffmanEncoder(REQUEST_CODES, REQUEST_CODES_LENGTH)
assert encoder.encode(b"www.example.com") == (
b'\xf1\xe3\xc2\xe5\xf2:k\xa0\xab\x90\xf4\xff')
assert encoder.encode(b"no-cache") == (b'\xa8\xeb\x10d\x9c\xbf')
assert encoder.encode(b"custom-key") == (b'%\xa8I\xe9[\xa9}\x7f')
assert encoder.encode(b"custom-value") == (
b'%\xa8I\xe9[\xb8\xe8\xb4\xbf')
def _load_pattern(self, verbose):
'''
Get query compressed size pattern.
'''
patterns = None
encoder = HuffmanEncoder(REQUEST_CODES, REQUEST_CODES_LENGTH)
def __get_patterns(string):
'''
Get possible size patterns of a query string.
Considering counter parameter if any exist.
'''
size_sequences = [[] for _ in range(8)]
# Initial bits number from 1 to 8 (+2 octets)
init_bytes = [
b'AA0', b'AAA', b'AAB', b'AAX', b'XX0', b'XXA', b'XXB', b'XXX'
]
query_bytes = b''
ct = self.website.ct_start
# Keep track of how far through an octet we are
for char in string:
if char == '\'' and self.website.encode_apost:
query_bytes += b'%27'
elif char == ' ' and self.website.encode_space:
query_bytes += b'%20'
else:
query_bytes += bytes(char, encoding='utf8')
if char == '\'':
if self.website.ct_type == 'cp':
ct += 1
continue
if char == ' ' and self.website.trim_space:
if self.website.ct_type == 'cp':
ct += 1
continue
if self.website.ct_type == 'no':
encode_bytes = query_bytes
else:
ct_bytes = bytes(str(ct), encoding='utf8')
encode_bytes = query_bytes + ct_bytes
for i in range(len(init_bytes)):
size_sequences[i].append(
len(encoder.encode(init_bytes[i] + encode_bytes)))
ct += 1
return [np.diff(x).tolist() for x in size_sequences]
# Compressed pattern is valid only for HTTP/2 requests without changing bytes
if self.website.http_version == 2 and self.website.change_byte == 0:
if verbose:
print('Loading query patterns')
if self.chinese:
patterns = self.__load_matedata('pattern_zh_%s.pkl' %
self.website)
if patterns is None:
tqdm.pandas(desc='pinyin')
string = self.queries['query'].progress_apply(
lambda x: '\''.join(py.get_pinyin(x).split('-')))
tqdm.pandas(desc='patterns')
patterns = string.progress_apply(__get_patterns)
self.__save_matedata(patterns,
'pattern_zh_%s.pkl' % self.website)
else:
patterns = self.__load_matedata('pattern_en_%s.pkl' %
self.website)
if patterns is None:
tqdm.pandas(desc='patterns')
patterns = self.queries['query'].progress_apply(
__get_patterns)
self.__save_matedata(patterns,
'pattern_en_%s.pkl' % self.website)
elif verbose:
if self.website.http_version == 1.1:
print('Compressed pattern is ignored for HTTP/1.1')
else:
print(
'Compressed pattern is invalid for changing byte parameters'
)
return patterns
def test_request_huffman_encode(self):
encoder = HuffmanEncoder(REQUEST_CODES, REQUEST_CODES_LENGTH)
assert encoder.encode(b"www.example.com") == (b'\xf1\xe3\xc2\xe5\xf2:k\xa0\xab\x90\xf4\xff')
assert encoder.encode(b"no-cache") == (b'\xa8\xeb\x10d\x9c\xbf')
assert encoder.encode(b"custom-key") == (b'%\xa8I\xe9[\xa9}\x7f')
assert encoder.encode(b"custom-value") == (b'%\xa8I\xe9[\xb8\xe8\xb4\xbf')