OccuRUtic;Nine

BaudotCodec.py

# -*- coding: utf-8 -*-


class BaudotCodec():

    encodes = {}
    decodes = {}
    ltr_trans_rule = {'保留域': 'NA',
                      'SPACE': ' ',
                      'LINE_FEED': '\n'}

    def __init__(self, filename='bc-list.txt'):
        with open(filename, 'r') as f:
            for line in f:
                code, ltr, fig = (line.strip().split(' ')+['dummy'])[:3]
                if ltr in self.ltr_trans_rule:
                    ltr = self.ltr_trans_rule[ltr]
                if fig in self.ltr_trans_rule:
                    fig = self.ltr_trans_rule[fig]
                elif fig == 'dummy':
                    fig = ltr

                self.encodes[ltr] = (code, False)
                self.encodes[fig] = (code, True)

        self.decodes = {v: k for k, v in self.encodes.items()}

    def encode(self, plaintext, bits='01'):
        wasfig = False
        codetext = ''

        for p in plaintext:
            code, isfig = self.encodes[p.upper()]
            if not wasfig and isfig:
                codetext += self.encodes['FIGS'][0]
            elif wasfig and not isfig:
                codetext += self.encodes['LTRS'][0]
            wasfig = isfig
            codetext += code

        if bits != '01':
            codetext = codetext.translate(str.maketrans('01', bits))

        return codetext

    def decode(self, codetext, bits='01'):
        isfig = False
        plaintext = ''

        if bits != '01':
            codetext = codetext.translate(str.maketrans(bits, '01'))

        for i in range(0, len(codetext), 5):
            code = codetext[i: i+5]
            if code == self.encodes['FIGS'][0]:
                isfig = True
            elif code == self.encodes['LTRS'][0]:
                isfig = False
            else:
                plaintext += self.decodes[(code, isfig)]

        return plaintext

bc-list.txt

00000 NULL
00001 E 3
00010 LINE_FEED
00011 A -
00100 SPACE
00101 S ' BELL
00110 I 8
00111 U 7
01000 CARRIAGE_RETURN
01001 D ENQ $
01010 R 4
01011 J BELL '
01100 N ,
01101 F 保留域 !
01110 C :
01111 K (
10000 T 5
10001 Z + "
10010 L )
10011 W 2
10100 H 保留域 STOP
10101 Y 6
10110 P 0
10111 Q 1
11000 O 9
11001 B ?
11010 G 保留域 &
11011 FIGS
11100 M .
11101 X /
11110 V ;
11111 LTRS