Innovenergy_trunk/firmware/Venus_Release/VenusReleaseFiles/pika-0.13.1/pika/data.py

323 lines
9.6 KiB
Python
Raw Normal View History

2024-07-01 09:49:57 +00:00
"""AMQP Table Encoding/Decoding"""
import struct
import decimal
import calendar
import warnings
from datetime import datetime
from pika import exceptions
from pika.compat import PY2, PY3
from pika.compat import unicode_type, long, as_bytes
def encode_short_string(pieces, value):
"""Encode a string value as short string and append it to pieces list
returning the size of the encoded value.
:param list pieces: Already encoded values
:param value: String value to encode
:type value: str or unicode
:rtype: int
"""
encoded_value = as_bytes(value)
length = len(encoded_value)
# 4.2.5.3
# Short strings, stored as an 8-bit unsigned integer length followed by zero
# or more octets of data. Short strings can carry up to 255 octets of UTF-8
# data, but may not contain binary zero octets.
# ...
# 4.2.5.5
# The server SHOULD validate field names and upon receiving an invalid field
# name, it SHOULD signal a connection exception with reply code 503 (syntax
# error).
# -> validate length (avoid truncated utf-8 / corrupted data), but skip null
# byte check.
if length > 255:
raise exceptions.ShortStringTooLong(encoded_value)
pieces.append(struct.pack('B', length))
pieces.append(encoded_value)
return 1 + length
if PY2:
def decode_short_string(encoded, offset):
"""Decode a short string value from ``encoded`` data at ``offset``.
"""
length = struct.unpack_from('B', encoded, offset)[0]
offset += 1
# Purely for compatibility with original python2 code. No idea what
# and why this does.
value = encoded[offset:offset + length]
try:
value = bytes(value)
except UnicodeEncodeError:
pass
offset += length
return value, offset
else:
def decode_short_string(encoded, offset):
"""Decode a short string value from ``encoded`` data at ``offset``.
"""
length = struct.unpack_from('B', encoded, offset)[0]
offset += 1
value = encoded[offset:offset + length]
try:
value = value.decode('utf8')
except UnicodeDecodeError:
pass
offset += length
return value, offset
def encode_table(pieces, table):
"""Encode a dict as an AMQP table appending the encded table to the
pieces list passed in.
:param list pieces: Already encoded frame pieces
:param dict table: The dict to encode
:rtype: int
"""
table = table or {}
length_index = len(pieces)
pieces.append(None) # placeholder
tablesize = 0
for (key, value) in table.items():
tablesize += encode_short_string(pieces, key)
tablesize += encode_value(pieces, value)
pieces[length_index] = struct.pack('>I', tablesize)
return tablesize + 4
def encode_value(pieces, value):
"""Encode the value passed in and append it to the pieces list returning
the the size of the encoded value.
:param list pieces: Already encoded values
:param any value: The value to encode
:rtype: int
"""
if PY2:
if isinstance(value, basestring):
if isinstance(value, unicode_type):
value = value.encode('utf-8')
pieces.append(struct.pack('>cI', b'S', len(value)))
pieces.append(value)
return 5 + len(value)
else:
# support only str on Python 3
if isinstance(value, str):
value = value.encode('utf-8')
pieces.append(struct.pack('>cI', b'S', len(value)))
pieces.append(value)
return 5 + len(value)
if isinstance(value, bytes):
pieces.append(struct.pack('>cI', b'x', len(value)))
pieces.append(value)
return 5 + len(value)
if isinstance(value, bool):
pieces.append(struct.pack('>cB', b't', int(value)))
return 2
if isinstance(value, long):
pieces.append(struct.pack('>cq', b'l', value))
return 9
elif isinstance(value, int):
with warnings.catch_warnings():
warnings.filterwarnings('error')
try:
p = struct.pack('>ci', b'I', value)
pieces.append(p)
return 5
except (struct.error, DeprecationWarning):
p = struct.pack('>cq', b'l', long(value))
pieces.append(p)
return 9
elif isinstance(value, decimal.Decimal):
value = value.normalize()
if value.as_tuple().exponent < 0:
decimals = -value.as_tuple().exponent
raw = int(value * (decimal.Decimal(10) ** decimals))
pieces.append(struct.pack('>cBi', b'D', decimals, raw))
else:
# per spec, the "decimals" octet is unsigned (!)
pieces.append(struct.pack('>cBi', b'D', 0, int(value)))
return 6
elif isinstance(value, datetime):
pieces.append(struct.pack('>cQ', b'T',
calendar.timegm(value.utctimetuple())))
return 9
elif isinstance(value, dict):
pieces.append(struct.pack('>c', b'F'))
return 1 + encode_table(pieces, value)
elif isinstance(value, list):
p = []
for v in value:
encode_value(p, v)
piece = b''.join(p)
pieces.append(struct.pack('>cI', b'A', len(piece)))
pieces.append(piece)
return 5 + len(piece)
elif value is None:
pieces.append(struct.pack('>c', b'V'))
return 1
else:
raise exceptions.UnsupportedAMQPFieldException(pieces, value)
def decode_table(encoded, offset):
"""Decode the AMQP table passed in from the encoded value returning the
decoded result and the number of bytes read plus the offset.
:param str encoded: The binary encoded data to decode
:param int offset: The starting byte offset
:rtype: tuple
"""
result = {}
tablesize = struct.unpack_from('>I', encoded, offset)[0]
offset += 4
limit = offset + tablesize
while offset < limit:
key, offset = decode_short_string(encoded, offset)
value, offset = decode_value(encoded, offset)
result[key] = value
return result, offset
def decode_value(encoded, offset):
"""Decode the value passed in returning the decoded value and the number
of bytes read in addition to the starting offset.
:param str encoded: The binary encoded data to decode
:param int offset: The starting byte offset
:rtype: tuple
:raises: pika.exceptions.InvalidFieldTypeException
"""
# slice to get bytes in Python 3 and str in Python 2
kind = encoded[offset:offset + 1]
offset += 1
# Bool
if kind == b't':
value = struct.unpack_from('>B', encoded, offset)[0]
value = bool(value)
offset += 1
# Short-Short Int
elif kind == b'b':
value = struct.unpack_from('>B', encoded, offset)[0]
offset += 1
# Short-Short Unsigned Int
elif kind == b'B':
value = struct.unpack_from('>b', encoded, offset)[0]
offset += 1
# Short Int
elif kind == b'U':
value = struct.unpack_from('>h', encoded, offset)[0]
offset += 2
# Short Unsigned Int
elif kind == b'u':
value = struct.unpack_from('>H', encoded, offset)[0]
offset += 2
# Long Int
elif kind == b'I':
value = struct.unpack_from('>i', encoded, offset)[0]
offset += 4
# Long Unsigned Int
elif kind == b'i':
value = struct.unpack_from('>I', encoded, offset)[0]
offset += 4
# Long-Long Int
elif kind == b'L':
value = long(struct.unpack_from('>q', encoded, offset)[0])
offset += 8
# Long-Long Unsigned Int
elif kind == b'l':
value = long(struct.unpack_from('>Q', encoded, offset)[0])
offset += 8
# Float
elif kind == b'f':
value = long(struct.unpack_from('>f', encoded, offset)[0])
offset += 4
# Double
elif kind == b'd':
value = long(struct.unpack_from('>d', encoded, offset)[0])
offset += 8
# Decimal
elif kind == b'D':
decimals = struct.unpack_from('B', encoded, offset)[0]
offset += 1
raw = struct.unpack_from('>i', encoded, offset)[0]
offset += 4
value = decimal.Decimal(raw) * (decimal.Decimal(10) ** -decimals)
# Short String
elif kind == b's':
value, offset = decode_short_string(encoded, offset)
# Long String
elif kind == b'S':
length = struct.unpack_from('>I', encoded, offset)[0]
offset += 4
value = encoded[offset:offset + length]
try:
value = value.decode('utf8')
except UnicodeDecodeError:
pass
offset += length
elif kind == b'x':
length = struct.unpack_from('>I', encoded, offset)[0]
offset += 4
value = encoded[offset:offset + length]
offset += length
# Field Array
elif kind == b'A':
length = struct.unpack_from('>I', encoded, offset)[0]
offset += 4
offset_end = offset + length
value = []
while offset < offset_end:
v, offset = decode_value(encoded, offset)
value.append(v)
# Timestamp
elif kind == b'T':
value = datetime.utcfromtimestamp(struct.unpack_from('>Q', encoded,
offset)[0])
offset += 8
# Field Table
elif kind == b'F':
(value, offset) = decode_table(encoded, offset)
# Null / Void
elif kind == b'V':
value = None
else:
raise exceptions.InvalidFieldTypeException(kind)
return value, offset