generated from mwc/lab_compression
47 lines
1.4 KiB
Python
47 lines
1.4 KiB
Python
from easybits import Bits
|
|
import codecs
|
|
|
|
def register_codec(encode, decode, name):
|
|
"""Registers a codec so that it can later be used to encode
|
|
or decode strings and bytes.
|
|
"""
|
|
def encode_wrapper(text):
|
|
return encode(text), len(text)
|
|
def decode_wrapper(data):
|
|
return decode(data), len(data)
|
|
def search_for_codec(query):
|
|
if query == name:
|
|
return codecs.CodecInfo(encode_wrapper, decode_wrapper, name=name)
|
|
codecs.register(search_for_codec)
|
|
|
|
com_neg= {"the": -1, "and": -2,
|
|
"this": -3, "or": -4, "not": -5, "that": -6,
|
|
"to": -7, "from": -8, "in": -9, "at": -10, "of": -11,
|
|
"because": -12, "but": -13, "be": -14,
|
|
"now": -15, "then": -16, "while": -17}
|
|
|
|
def common_encode(commonword):
|
|
return com_neg.get(commonword)
|
|
|
|
def encode(text):
|
|
result = Bits()
|
|
for char in text:
|
|
if text in com_neg:
|
|
result= common_encode(text)
|
|
else:
|
|
try:
|
|
b = Bits(char, encoding='ascii')
|
|
except UnicodeEncodeError:
|
|
b = Bits('?')
|
|
result = result.concat(b[1:])
|
|
return result.bytes
|
|
|
|
def decode(data):
|
|
bits = Bits(bytes(data))
|
|
text = ""
|
|
for i in range(-17, len(bits), 7):
|
|
byte = Bits('0').concat(bits[i:i+7])
|
|
text += Bits(byte).ascii
|
|
return text
|
|
|
|
register_codec(encode, decode, "ascii7") |