/usr/lib/python2.7/dist-packages/hpack/hpack.py is in python-hpack 2.3.0-1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 | # -*- coding: utf-8 -*-
"""
hpack/hpack
~~~~~~~~~~~
Implements the HPACK header compression algorithm as detailed by the IETF.
"""
import logging
from .table import HeaderTable, table_entry_size
from .compat import to_byte, to_bytes
from .exceptions import HPACKDecodingError, OversizedHeaderListError
from .huffman import HuffmanEncoder
from .huffman_constants import (
REQUEST_CODES, REQUEST_CODES_LENGTH
)
from .huffman_table import decode_huffman
from .struct import HeaderTuple, NeverIndexedHeaderTuple
log = logging.getLogger(__name__)
INDEX_NONE = b'\x00'
INDEX_NEVER = b'\x10'
INDEX_INCREMENTAL = b'\x40'
try: # pragma: no cover
basestring = basestring
except NameError: # pragma: no cover
basestring = (str, bytes)
# We default the maximum header list we're willing to accept to 64kB. That's a
# lot of headers, but if applications want to raise it they can do.
DEFAULT_MAX_HEADER_LIST_SIZE = 2**16
def _unicode_if_needed(header, raw):
"""
Provides a header as a unicode string if raw is False, otherwise returns
it as a bytestring.
"""
name = to_bytes(header[0])
value = to_bytes(header[1])
if not raw:
name = name.decode('utf-8')
value = value.decode('utf-8')
return header.__class__(name, value)
def encode_integer(integer, prefix_bits):
"""
This encodes an integer according to the wacky integer encoding rules
defined in the HPACK spec.
"""
log.debug("Encoding %d with %d bits", integer, prefix_bits)
max_number = (2 ** prefix_bits) - 1
if integer < max_number:
return bytearray([integer]) # Seriously?
else:
elements = [max_number]
integer -= max_number
while integer >= 128:
elements.append((integer % 128) + 128)
integer //= 128 # We need integer division
elements.append(integer)
return bytearray(elements)
def decode_integer(data, prefix_bits):
"""
This decodes an integer according to the wacky integer encoding rules
defined in the HPACK spec. Returns a tuple of the decoded integer and the
number of bytes that were consumed from ``data`` in order to get that
integer.
"""
max_number = (2 ** prefix_bits) - 1
mask = 0xFF >> (8 - prefix_bits)
index = 0
try:
number = to_byte(data[index]) & mask
if number == max_number:
while True:
index += 1
next_byte = to_byte(data[index])
# There's some duplication here, but that's because this is a
# hot function, and incurring too many function calls here is
# a real problem. For that reason, we unrolled the maths.
if next_byte >= 128:
number += (next_byte - 128) * (128 ** (index - 1))
else:
number += next_byte * (128 ** (index - 1))
break
except IndexError:
raise HPACKDecodingError(
"Unable to decode HPACK integer representation from %r" % data
)
log.debug("Decoded %d, consumed %d bytes", number, index + 1)
return number, index + 1
def _dict_to_iterable(header_dict):
"""
This converts a dictionary to an iterable of two-tuples. This is a
HPACK-specific function becuase it pulls "special-headers" out first and
then emits them.
"""
assert isinstance(header_dict, dict)
keys = sorted(
header_dict.keys(),
key=lambda k: not _to_bytes(k).startswith(b':')
)
for key in keys:
yield key, header_dict[key]
def _to_bytes(string):
"""
Convert string to bytes.
"""
if not isinstance(string, (basestring)): # pragma: no cover
string = str(string)
return string if isinstance(string, bytes) else string.encode('utf-8')
class Encoder(object):
"""
An HPACK encoder object. This object takes HTTP headers and emits encoded
HTTP/2 header blocks.
"""
def __init__(self):
self.header_table = HeaderTable()
self.huffman_coder = HuffmanEncoder(
REQUEST_CODES, REQUEST_CODES_LENGTH
)
self.table_size_changes = []
@property
def header_table_size(self):
"""
Controls the size of the HPACK header table.
"""
return self.header_table.maxsize
@header_table_size.setter
def header_table_size(self, value):
self.header_table.maxsize = value
if self.header_table.resized:
self.table_size_changes.append(value)
def encode(self, headers, huffman=True):
"""
Takes a set of headers and encodes them into a HPACK-encoded header
block.
:param headers: The headers to encode. Must be either an iterable of
tuples, an iterable of :class:`HeaderTuple
<hpack.struct.HeaderTuple>`, or a ``dict``.
If an iterable of tuples, the tuples may be either
two-tuples or three-tuples. If they are two-tuples, the
tuples must be of the format ``(name, value)``. If they
are three-tuples, they must be of the format
``(name, value, sensitive)``, where ``sensitive`` is a
boolean value indicating whether the header should be
added to header tables anywhere. If not present,
``sensitive`` defaults to ``False``.
If an iterable of :class:`HeaderTuple
<hpack.struct.HeaderTuple>`, the tuples must always be
two-tuples. Instead of using ``sensitive`` as a third
tuple entry, use :class:`NeverIndexedHeaderTuple
<hpack.struct.NeverIndexedHeaderTuple>` to request that
the field never be indexed.
.. warning:: HTTP/2 requires that all special headers
(headers whose names begin with ``:`` characters)
appear at the *start* of the header block. While
this method will ensure that happens for ``dict``
subclasses, callers using any other iterable of
tuples **must** ensure they place their special
headers at the start of the iterable.
For efficiency reasons users should prefer to use
iterables of two-tuples: fixing the ordering of
dictionary headers is an expensive operation that
should be avoided if possible.
:param huffman: (optional) Whether to Huffman-encode any header sent as
a literal value. Except for use when debugging, it is
recommended that this be left enabled.
:returns: A bytestring containing the HPACK-encoded header block.
"""
# Transforming the headers into a header block is a procedure that can
# be modeled as a chain or pipe. First, the headers are encoded. This
# encoding can be done a number of ways. If the header name-value pair
# are already in the header table we can represent them using the
# indexed representation: the same is true if they are in the static
# table. Otherwise, a literal representation will be used.
log.debug("HPACK encoding %s", headers)
header_block = []
# Turn the headers into a list of tuples if possible. This is the
# natural way to interact with them in HPACK. Because dictionaries are
# un-ordered, we need to make sure we grab the "special" headers first.
if isinstance(headers, dict):
headers = _dict_to_iterable(headers)
# Before we begin, if the header table size has been changed we need
# to signal all changes since last emission appropriately.
if self.header_table.resized:
header_block.append(self._encode_table_size_change())
self.header_table.resized = False
# Add each header to the header block
for header in headers:
sensitive = False
if isinstance(header, HeaderTuple):
sensitive = not header.indexable
elif len(header) > 2:
sensitive = header[2]
header = (_to_bytes(header[0]), _to_bytes(header[1]))
header_block.append(self.add(header, sensitive, huffman))
header_block = b''.join(header_block)
log.debug("Encoded header block to %s", header_block)
return header_block
def add(self, to_add, sensitive, huffman=False):
"""
This function takes a header key-value tuple and serializes it.
"""
log.debug("Adding %s to the header table", to_add)
name, value = to_add
# Set our indexing mode
indexbit = INDEX_INCREMENTAL if not sensitive else INDEX_NEVER
# Search for a matching header in the header table.
match = self.header_table.search(name, value)
if match is None:
# Not in the header table. Encode using the literal syntax,
# and add it to the header table.
encoded = self._encode_literal(name, value, indexbit, huffman)
if not sensitive:
self.header_table.add(name, value)
return encoded
# The header is in the table, break out the values. If we matched
# perfectly, we can use the indexed representation: otherwise we
# can use the indexed literal.
index, name, perfect = match
if perfect:
# Indexed representation.
encoded = self._encode_indexed(index)
else:
# Indexed literal. We are going to add header to the
# header table unconditionally. It is a future todo to
# filter out headers which are known to be ineffective for
# indexing since they just take space in the table and
# pushed out other valuable headers.
encoded = self._encode_indexed_literal(
index, value, indexbit, huffman
)
if not sensitive:
self.header_table.add(name, value)
return encoded
def _encode_indexed(self, index):
"""
Encodes a header using the indexed representation.
"""
field = encode_integer(index, 7)
field[0] |= 0x80 # we set the top bit
return bytes(field)
def _encode_literal(self, name, value, indexbit, huffman=False):
"""
Encodes a header with a literal name and literal value. If ``indexing``
is True, the header will be added to the header table: otherwise it
will not.
"""
if huffman:
name = self.huffman_coder.encode(name)
value = self.huffman_coder.encode(value)
name_len = encode_integer(len(name), 7)
value_len = encode_integer(len(value), 7)
if huffman:
name_len[0] |= 0x80
value_len[0] |= 0x80
return b''.join(
[indexbit, bytes(name_len), name, bytes(value_len), value]
)
def _encode_indexed_literal(self, index, value, indexbit, huffman=False):
"""
Encodes a header with an indexed name and a literal value and performs
incremental indexing.
"""
if indexbit != INDEX_INCREMENTAL:
prefix = encode_integer(index, 4)
else:
prefix = encode_integer(index, 6)
prefix[0] |= ord(indexbit)
if huffman:
value = self.huffman_coder.encode(value)
value_len = encode_integer(len(value), 7)
if huffman:
value_len[0] |= 0x80
return b''.join([bytes(prefix), bytes(value_len), value])
def _encode_table_size_change(self):
"""
Produces the encoded form of all header table size change context
updates.
"""
block = b''
for size_bytes in self.table_size_changes:
size_bytes = encode_integer(size_bytes, 5)
size_bytes[0] |= 0x20
block += bytes(size_bytes)
self.table_size_changes = []
return block
class Decoder(object):
"""
An HPACK decoder object.
.. versionchanged:: 2.3.0
Added ``max_header_list_size`` argument.
:param max_header_list_size: The maximum decompressed size we will allow
for any single header block. This is a protection against DoS attacks
that attempt to force the application to expand a relatively small
amount of data into a really large header list, allowing enormous
amounts of memory to be allocated.
If this amount of data is exceeded, a `OversizedHeaderListError
<hpack.OversizedHeaderListError>` exception will be raised. At this
point the connection should be shut down, as the HPACK state will no
longer be useable.
Defaults to 64kB.
:type max_header_list_size: ``int``
"""
def __init__(self, max_header_list_size=DEFAULT_MAX_HEADER_LIST_SIZE):
self.header_table = HeaderTable()
#: The maximum decompressed size we will allow for any single header
#: block. This is a protection against DoS attacks that attempt to
#: force the application to expand a relatively small amount of data
#: into a really large header list, allowing enormous amounts of memory
#: to be allocated.
#:
#: If this amount of data is exceeded, a `OversizedHeaderListError
#: <hpack.OversizedHeaderListError>` exception will be raised. At this
#: point the connection should be shut down, as the HPACK state will no
#: longer be useable.
#:
#: Defaults to 64kB.
#:
#: .. versionadded:: 2.3.0
self.max_header_list_size = max_header_list_size
@property
def header_table_size(self):
"""
Controls the size of the HPACK header table.
"""
return self.header_table.maxsize
@header_table_size.setter
def header_table_size(self, value):
self.header_table.maxsize = value
def decode(self, data, raw=False):
"""
Takes an HPACK-encoded header block and decodes it into a header set.
:param data: A bytestring representing a complete HPACK-encoded header
block.
:param raw: (optional) Whether to return the headers as tuples of raw
byte strings or to decode them as UTF-8 before returning
them. The default value is False, which returns tuples of
Unicode strings
:returns: A list of two-tuples of ``(name, value)`` representing the
HPACK-encoded headers, in the order they were decoded.
:raises HPACKDecodingError: If an error is encountered while decoding
the header block.
"""
log.debug("Decoding %s", data)
data_mem = memoryview(data)
headers = []
data_len = len(data)
inflated_size = 0
current_index = 0
while current_index < data_len:
# Work out what kind of header we're decoding.
# If the high bit is 1, it's an indexed field.
current = to_byte(data[current_index])
indexed = bool(current & 0x80)
# Otherwise, if the second-highest bit is 1 it's a field that does
# alter the header table.
literal_index = bool(current & 0x40)
# Otherwise, if the third-highest bit is 1 it's an encoding context
# update.
encoding_update = bool(current & 0x20)
if indexed:
header, consumed = self._decode_indexed(
data_mem[current_index:]
)
elif literal_index:
# It's a literal header that does affect the header table.
header, consumed = self._decode_literal_index(
data_mem[current_index:]
)
elif encoding_update:
# It's an update to the encoding context.
consumed = self._update_encoding_context(data_mem)
header = None
else:
# It's a literal header that does not affect the header table.
header, consumed = self._decode_literal_no_index(
data_mem[current_index:]
)
if header:
headers.append(header)
inflated_size += table_entry_size(*header)
if inflated_size > self.max_header_list_size:
raise OversizedHeaderListError(
"A header list larger than %d has been received" %
self.max_header_list_size
)
current_index += consumed
try:
return [_unicode_if_needed(h, raw) for h in headers]
except UnicodeDecodeError:
raise HPACKDecodingError("Unable to decode headers as UTF-8.")
def _update_encoding_context(self, data):
"""
Handles a byte that updates the encoding context.
"""
# We've been asked to resize the header table.
new_size, consumed = decode_integer(data, 5)
self.header_table_size = new_size
return consumed
def _decode_indexed(self, data):
"""
Decodes a header represented using the indexed representation.
"""
index, consumed = decode_integer(data, 7)
header = HeaderTuple(*self.header_table.get_by_index(index))
log.debug("Decoded %s, consumed %d", header, consumed)
return header, consumed
def _decode_literal_no_index(self, data):
return self._decode_literal(data, False)
def _decode_literal_index(self, data):
return self._decode_literal(data, True)
def _decode_literal(self, data, should_index):
"""
Decodes a header represented with a literal.
"""
total_consumed = 0
# When should_index is true, if the low six bits of the first byte are
# nonzero, the header name is indexed.
# When should_index is false, if the low four bits of the first byte
# are nonzero the header name is indexed.
if should_index:
indexed_name = to_byte(data[0]) & 0x3F
name_len = 6
not_indexable = False
else:
high_byte = to_byte(data[0])
indexed_name = high_byte & 0x0F
name_len = 4
not_indexable = high_byte & 0x10
if indexed_name:
# Indexed header name.
index, consumed = decode_integer(data, name_len)
name = self.header_table.get_by_index(index)[0]
total_consumed = consumed
length = 0
else:
# Literal header name. The first byte was consumed, so we need to
# move forward.
data = data[1:]
length, consumed = decode_integer(data, 7)
name = data[consumed:consumed + length]
if to_byte(data[0]) & 0x80:
name = decode_huffman(name)
total_consumed = consumed + length + 1 # Since we moved forward 1.
data = data[consumed + length:]
# The header value is definitely length-based.
length, consumed = decode_integer(data, 7)
value = data[consumed:consumed + length]
if to_byte(data[0]) & 0x80:
value = decode_huffman(value)
# Updated the total consumed length.
total_consumed += length + consumed
# If we have been told never to index the header field, encode that in
# the tuple we use.
if not_indexable:
header = NeverIndexedHeaderTuple(name, value)
else:
header = HeaderTuple(name, value)
# If we've been asked to index this, add it to the header table.
if should_index:
self.header_table.add(name, value)
log.debug(
"Decoded %s, total consumed %d bytes, indexed %s",
header,
total_consumed,
should_index
)
return header, total_consumed
|