pocketbook2tsv.py

import os
import re
import sys
import zlib

from struct import unpack

OFFSET = 0x6B4

"""
Zlib streams:
    0: Alphabet?
    1: Language, Keyboard
    2: Wordlist
    3+: Definitions
"""

def convert(fname: str) -> None:
    defilist: list[tuple[str, str]] = []
    info = ""
    j = 0
    i = 0
    with open(fname, "rb") as f:
        dic = f.read() # [OFFSET:]
    while j < len(dic):
        try:
            decomp_obj = zlib.decompressobj()
            decomp = decomp_obj.decompress(dic[j:])
            try:
                if i == 0:
                    decoded = decomp.decode("utf-16")
                    decoded = "\n".join(decoded.split("\x00"))
                    info += f"{decoded}\n\n"
                elif i == 1:
                    decoded = decomp.decode()
                    info += f"{decoded}\n\n"
                elif i == 2:
                    lst = decomp.split(b"\x00")
                    decoded = []
                    for word in lst:
                        try:
                            word_decoded = word[2:].decode()
                            if word_decoded:
                                decoded.append(word_decoded)
                        except:
                            pass
                    with_endline = '\n'.join(decoded)
                    info += f"Length of Wordlist: {len(decoded)}\n{with_endline}"
                else:
                    pos = 0
                    while True:
                        _len = unpack("<H", decomp[pos:pos+2])[0]
                        if _len == 0:
                            break
                        idx = (pair := decomp[pos+2:pos+_len]).index(b"\x00")
                        hw = pair[0:idx].decode()
                        defi = pair[idx+1:].decode().replace("\n","\\n").removesuffix("\\n")
                        defi = re.sub("\s{2,}", " ", defi)
                        defi = re.sub("[\x00-\x1F]+", "", defi)
                        defilist.append((hw, defi))
                        pos += _len
                i += 1
            except Exception as e:
                print(e)
            j += len(dic[j:]) - len(decomp_obj.unused_data)
        except zlib.error:
            j += 1

    outname = os.path.splitext(os.path.basename(fname))[0]
    with open(f"{outname}.tsv", "w", encoding="utf-8") as tsv:
        for i in defilist:
            tsv.write(f"{i[0]}\t{i[1]}\n")
    with open(f"{outname}_info.txt", "w", encoding="utf-8") as txt:
        txt.write(info)

if __name__ == '__main__':
    if len(sys.argv) < 2:
        sys.exit("Missing filename.")
    convert(sys.argv[1])