《蓓优妮塔3》PTD文本文件分析笔记

#!/usr/bin/python3
# -*- coding: UTF-8 -*-

import io
import os
import struct
import json

def decode_string(byte_array):
  byte_array = list(map(lambda x: (x - 0x26 + 0x100) % 0x100, byte_array))
  assert byte_array[-2:] == [0, 0]
  return bytes(byte_array[:-2]).decode("utf-16-le", "replace")

def hash_format(f):
  _hash, = struct.unpack("<I", f.read(4))
  return f"{_hash:08x}"

def export_ptd(f: io.BufferedReader):
  assert f.read(4) == b"PTD\0"
  _1, _2, hash_count, hash_data_pos, _4, string_data_pos = struct.unpack("<6I", f.read(24))
  assert _1 == 0x02
  assert _2 == 0x26
  assert _4 == 0x01
  data = {
    "_1": _1,
    "_2": _2,
    "hash_count": hash_count,
    "hash_data_pos": hash_data_pos,
    "_4": _4,
    "string_data_pos": string_data_pos,
    "hash_table": {},
    "texts": {}
  }

  # hash 表
  hash_name_pos = hash_data_pos + hash_count * 0x10
  for i in range(hash_count):
    f.seek(hash_data_pos + i * 0x10)
    hash = hash_format(f)
    name_pos, char_length, byte_length = struct.unpack("<3I", f.read(12))
    assert byte_length == char_length * 2
    f.seek(hash_name_pos)
    name = decode_string(f.read(byte_length))
    hash_name_pos = f.tell()
    data["hash_table"][hash] = name
  assert string_data_pos == f.tell()
  # assert "3b8ba7c7" in data["hash_table"]
  # assert data["hash_table"]["3b8ba7c7"]["name"] == "Text"

  # 文件信息
  hash = hash_format(f)
  has_groupid, __2, __3, text_pos = struct.unpack("<4I", f.read(16))
  assert has_groupid in [0, 1]
  assert __2 == 0x14
  assert __3 == 0x02

  # groupid
  if "7805ac12" in data["hash_table"]:
    hash = hash_format(f)
    assert hash == "7805ac12"
    groupid_count, __2 = struct.unpack("<2I", f.read(8))
    assert __2 == 0x0c
    groupids = list(map(lambda x: data["hash_table"][hash_format(f)], range(groupid_count)))
    # data["groupids"] = groupids
  else:
    # 没有 groupid，无文本
    assert has_groupid == 0
    pass
  
  # Text
  hash = hash_format(f)
  assert hash == "3b8ba7c7"
  text_count, __2 = struct.unpack("<2I", f.read(8))
  if has_groupid == 0:
    assert text_count == 0
  else:
    assert text_count > 0
    assert groupid_count == text_count
  assert __2 == 0x18

  # CharName
  hash = hash_format(f)
  assert hash == "3cc2b4f9"
  __1, __2 = struct.unpack("<2I", f.read(8))
  assert __1 == text_count

  text_data_pos = f.tell()
  text_pos = text_data_pos + text_count * 0x10
  for i in range(text_count):
    f.seek(text_data_pos + i * 0x10)
    hash = hash_format(f)
    name_pos, char_length, byte_length = struct.unpack("<3I", f.read(12))
    assert byte_length == char_length * 2
    f.seek(text_pos)
    text = decode_string(f.read(byte_length))
    text_pos = f.tell()
    data["texts"][data["hash_table"][hash]] = text
  
  if has_groupid == 1:
    data["texts"] = {
      k: data["texts"][k] for k in sorted(data["texts"].keys(), key=groupids.index)
    }
  return data

if __name__ == "__main__":
  writers = { lang: open(f"{lang}.txt", "w", -1, "utf8") for lang in ["CNzh", "TWzh", "USen", "JPja"] }

  for root, dirs, files in os.walk("Text_000"):
    for file in files:
      with open(os.path.join(root, file), "rb") as f:
        if not file.endswith(".bin"):
          continue

        data = export_ptd(f)

      with open(os.path.join(root, file.replace(".bin", ".json")), "w", -1, "utf8") as f:
        json.dump(data, f, ensure_ascii=False, indent=2)

      file_path = "/".join(os.path.join(root, file).replace("\\", "/").split("/")[-2:])
      *_, lang = file_path.split(".")[0].split("_")
      file_name = "_".join(_)
      if lang in writers:
        writers[lang].write(f"{file_name}\n")
        for name, text in data["texts"].items():
          text = text.replace("\t", "\\t").replace("\r\n", "\\n").replace("\n", "\\n")
          writers[lang].write(f"{name}\t{text}\n")

  for lang in writers:
    writers[lang].close()