from math import log from gzip import compress, decompress MAP = [] # CJK (Chinese characters) ranges = [range(0x4E00, 0x9FFF), range(0x3400, 0x4DBF), range(0x20000, 0x2A6DF), range(0x2A700, 0x2B73F), range(0x2B740, 0x2B81F), range(0x2B820, 0x2CEAF), range(0x2CEB0, 0x2EBEF), range(0x30000, 0x3134F)] MAP += [chr(x) for y in ranges for x in y] # Some Emoticons #MAP += [chr(x) for x in range(0x1F600, 0x1F64F)] BASE = int(log(len(MAP)+1, 2)) GZIP=True def encode(input, gzip=False): if gzip: result = compress(input.encode()) result = [bin(int(x))[2:] for x in result] else: result = [bin(ord(x))[2:] for x in input] result = ["0"*(8 - len(s)) + s for s in result] result = "".join(result) result = [result[i:i+BASE] for i in range(0, len(result), BASE)] result = [s + "0"*(BASE - len(s)) for s in result] result = [int(x,2) for x in result] result = [MAP[x] for x in result] return "".join(result) def decode(input, gzip=False): result = [MAP.index(x) for x in input] if gzip: result = [bin(x)[2:] for x in result] else: result = [MAP.index(x) for x in input] result = [bin(x)[2:] for x in result if not x==0] result = ["0"*(BASE - len(s)) + s for s in result] result = "".join(result) result = [result[i:i+8] for i in range(0, len(result), 8)] result = [s + "0"*(8 - len(s)) for s in result] if gzip: result = bytes([int(x, 2) for x in result]) return decompress(result).decode() result = [chr(int(x, 2)) for x in result if not int(x, 2)==0] return "".join(result) def test(): text = ( 'This is a longer text encoded in CJK characters. This way,' ' messages of more than 150 European characters can be sent through' ' Twitter. Is\'t that great!? To add more characters in this demo:' ' äÖüéß@%' '\n[-space for more chars-]' '\nWith gzip enabled, even >8-bit Unicode characters are supported:' ' € (EUR)' ) #print("INPUT:", "\n"+text, "\n(", len(text), "chars )") encoded = encode(text, GZIP) #print("ENCODED:", "\n"+encoded, "\n(", len(encoded), "chars )") decoded = decode(encoded, GZIP) #print("DECODED:", "\n"+decoded, "\n(", len(decoded), "chars )") assert text == decode(encode(text, True), True) if __name__ == "__main__": test() text = input("Reading message from input: ") print("INPUT:", "\n"+text, "\n(", len(text), "chars )") decoded = decode(text, GZIP) print("DECODED:", "\n"+decoded, "\n(", len(decoded), "chars )")