环境工具
编码环境：Jupyter、python、vscode
游戏样本：Eclipse no majo
游戏解包工具: Assets studio、UnityPy、uabead
游戏资源分析
资源文件目录分析
AssetBundle目录的结构如下:
-AssetBundle
-- Japanese
--- lc_tables
--- scenario
-- English
--- lc_tables
--- scenario
-- ...可见有Japanese和English两个分语言的文件夹，可以推断游戏有采取i18n方案。
在Assets Studio浏览i18n游戏资源发现，English文本不完善，故而选择进一步分析Japanese文件夹下的资源文件
文件资源分析
主要分析的是两个文件：lc_tables和scenario
lc_tables
存放的是主要为游戏的基础信息，如道具名称，结局名称，格式为json。
scenario
存放的是游戏的对话场景台本，需要翻译的内容大致分为以下几类:
- 人物台词：以“「”开头以“」”结尾
- 旁白：同人物台词，但是没有“「”和“」”作为前导或尾随字符
- 功能文本：形如：\[A-Z]{2}["*+"]
翻译过程
- 使用UnityPy对游戏资源解包
- 翻译匹配形式的文本
- 重新封包，需要使用到uabead重新压缩
完整代码
提取资源
python
import os
import UnityPy
def extract_TextAsset(game_path: str, output_path: str = "output"):
"""提取Texture2D"""
for root, dirs, files in os.walk(game_path):
for file_name in files:
asset_file_path = os.path.join(root, file_name)
env = UnityPy.load(asset_file_path)
for obj in env.objects:
if(obj.type.name == "TextAsset"):
path_id = obj.path_id
data = obj.read()
container_path = str(obj.container)
data_name = container_path[container_path.rindex("/")+1:container_path.rindex(".")]
ext = container_path[container_path.rindex(".")+1:]
with open(os.path.join(output_path ,f"{file_name}_{data_name}_{path_id}.{ext}"), "wb+") as f:
f.write(bytes(data.script))
extract_TextAsset("game\AssetBundles", "raw")翻译
python
import os
import re
import json
from hashlib import md5
import random
import requests
import time
def baidu_translate(content: str) -> str|None:
"""基于requests的百度翻译"""
url = "https://api.fanyi.baidu.com/api/trans/vip/translate"
q = content
from_lang = "jp"
to_lang = "zh"
appid = "******"
secret = "******"
salt = str(random.randint(100000,999999))
sign = md5((appid + q + salt + secret).encode()).hexdigest()
data = {
"q": q,
"from": from_lang,
"to": to_lang,
"appid": appid,
"salt": salt,
"sign": sign
}
try:
resp = requests.post(url, data, headers={
"Content-Type": "application/x-www-form-urlencoded"
})
except Exception as e:
print(e)
return None
if resp.status_code != 200:
print("Http请求错误!")
return None
return "".join(result['dst'] + '\n' for result in resp.json()['trans_result'])[0:-1]
class TextAssetsTranslator:
def __init__(self) -> None:
self.charaters: dict[str, str] = {"リズ":"丽兹","エリ":"艾丽","ロクサーヌ":"罗克珊"}
# def test(self, content: str, start_and_end: str):
# return start_and_end + content + start_and_end
def send_translate_request(self, content: str) -> str:
time.sleep(.1)
content = content.replace("<color=#f59aa0>♥</color>", "♥").replace("\V[EnemyName]", "EnemyName")
result = baidu_translate(content)
if result == None:
time.sleep(.1)
result = baidu_translate(content)
if result == None:
raise Exception("百度翻译请求失败！")
result = result.replace("♥","<color=#f59aa0>♥</color>").replace("EnemyName", "\V[EnemyName]")
origin_length = len(content.split("\n"))
result_length = len(result.split("\n"))
result = result + "\n" * (origin_length - result_length)
return result
def translate_scenarios(self, raw_text_folder_path: str = "raw", output_path: str = "translated", match_pattern: str|None = None):
"""批量翻译场景文件"""
for item in os.scandir(raw_text_folder_path):
if item.is_dir():
continue
file_name = item.name
# Skip when file name don't match the pattern
if match_pattern != None:
if re.match(match_pattern, file_name) == None:
continue
# Skip when file is already exist
if file_name in os.listdir(output_path):
continue
raw_file_path = os.path.join(raw_text_folder_path, file_name)
print(f"loading {raw_file_path}\n")
# Read data
with open(raw_file_path, "r", encoding="utf-8") as f:
line_list = f.readlines()
# Edit data by line
read_index = 0
while read_index < len(line_list):
line = line_list[read_index]
# Skip no japanese content line
if re.search(r"([^\x00-\xff])+", line) == None:
read_index = read_index + 1
continue
# Skip comment
if re.match(r"^\s*//", line):
read_index = read_index + 1
continue
# Skip logical syntax
if re.match(r"^\s*:", line):
read_index = read_index + 1
continue
# Handle charater name
if re.match(r"\s*\\CN\[.+\]",line):
raw_charater_name = line[line.find('"') + 1:line.rindex('"')]
if self.charaters.get(raw_charater_name) == None:
charater_name = self.send_translate_request(raw_charater_name)
self.charaters[raw_charater_name] = charater_name
print(f"[角色名]{raw_charater_name} => {charater_name}")
else:
charater_name = self.charaters[raw_charater_name]
# Edit
line = line.replace(raw_charater_name, charater_name)
# Save edit
line_list[read_index] = line
# Handle choice
if re.match(r"\s*\\CI\[.+\]",line):
raw = line[line.find('"') + 1:line.rindex('"')]
result = self.send_translate_request(raw)
print(f"[选项]{raw} => {result}")
# Edit
line = line.replace(raw, result)
# Save edit
line_list[read_index] = line
# Handle dialog
elif re.match(r"^\s*「([^」]+」\s*$|[^」]+$)", line):
start = read_index
raw_contents = [line[line.find("「") + 1:]]
while re.match(r".+」$", line) == None:
read_index = read_index + 1
line = line_list[read_index]
raw_contents.append(line)
raw_contents[-1] = raw_contents[-1][:raw_contents[-1].rindex("」")]
# Handle
wrapped_content = "".join(raw_contents)
handled_content = self.send_translate_request(wrapped_content)
result_contents = [content + "\n" for content in handled_content.split("\n")]
result_contents[-1] = result_contents[-1][:-1]
# Edit
for i in range(len(raw_contents)):
raw = raw_contents[i].replace("\n","").replace(" ", "").replace("\t", "").replace(" ","")
result = result_contents[i].replace("\n","").replace(" ","").replace("\t", "").replace(" ","")
line_list[start + i] = line_list[start + i].replace(raw, result)
print(f"[对话]{raw} => {result}")
# Handle voice-over
elif re.match(r"^\s*(\\V\[EnemyName\])?[^\x00-\xff]+", line):
start = read_index
raw_contents: list[str] = []
while re.match(r"^\s*(\\V\[EnemyName\])?[^\x00-\xff]+", line) and re.match(r"^\s*「([^」]+」\s*$|[^」]+$)", line) == None:
raw_contents.append(line)
read_index = read_index + 1
if read_index >= len(line_list): break
line = line_list[read_index]
raw_contents[-1] = raw_contents[-1][:-1]
# Handle
wrapped_content = "".join(raw_contents)
handled_content = self.send_translate_request(wrapped_content)
result_contents = [content + "\n" for content in handled_content.split("\n")]
# Edit
for i in range(len(raw_contents)):
raw = raw_contents[i].replace("\n","").replace(" ","").replace("\t","").replace(" ","")
result = result_contents[i].replace("\n","").replace(" ","").replace("\t", "").replace(" ","")
line_list[start + i] = line_list[start + i].replace(raw, result)
print(f"[旁白]{raw} => {result}")
read_index = read_index + 1
# Write data
output_file_path = os.path.join(output_path, file_name)
with open(output_file_path, "w+", encoding="utf-8") as f:
f.write("".join(line_list))
def translate_json(self, raw_text_folder_path: str = "raw", output_path: str = "translated", match_pattern: str|None = None):
"""翻译json"""
for item in os.scandir(raw_text_folder_path):
if item.is_dir():
continue
file_name = item.name
# Skip when file name don't match the pattern
if match_pattern != None:
if re.match(match_pattern, file_name) == None:
continue
# Skip when file is already exist
if file_name in os.listdir(output_path):
continue
json_file_path = os.path.join(raw_text_folder_path, file_name)
print(f"loading {json_file_path}")
with open(json_file_path, "r", encoding="utf-8") as f:
str_data = f.read()
data = json.loads(str_data)
stack: list[dict|list|bool|int|float|str] = [data]
trans_dict: dict[str,str] = {}
while len(stack) > 0:
current_node = stack.pop()
if type(current_node) is dict:
for key, value in current_node.items():
if type(value) is str:
if re.search(r"[^\x00-\xff◯×]+", value):
# Translate
if trans_dict.get(value) == None:
trans_dict[value] = self.send_translate_request(value)
print(f"{value} => {trans_dict[value]}")
# Edit
current_node[key] = trans_dict[value]
else:
stack.append(value)
if type(current_node) is list:
for i, item in enumerate(current_node):
if type(item) is str:
if re.search(r"[^\x00-\xff◯×]+", item):
# Translate
if trans_dict.get(item) == None:
trans_dict[item] = self.send_translate_request(item)
print(f"{item} => {trans_dict[item]}")
# Edit
current_node[i] = trans_dict[item]
else:
stack.append(item)
# Save
output_file_path = os.path.join(output_path, file_name)
with open(output_file_path, "w+", encoding="utf-8") as f:
str_data = json.dumps(data)
f.write(str_data)
translator = TextAssetsTranslator()
translator.translate_scenarios(match_pattern=r"scenarios_.+\.txt")
translator.translate_json(match_pattern=r"(lc_)?tables_.+\.json")打包
python
import os
import UnityPy
def inject_TextAsset(game_path: str, source_path: str = "translated", output_path: str = "output", match_pattern: str|None = None):
"""写入文本"""
source_files = os.listdir(source_path)
for root, dirs, files in os.walk(game_path):
for file_name in files:
asset_file_path = os.path.join(root, file_name)
env = UnityPy.load(asset_file_path)
change_flag = False
for container_path, obj in env.container.items():
if obj.type.name == "TextAsset":
path_id = obj.path_id
asset_name = container_path[container_path.rindex("/") + 1 : container_path.rindex(".")]
ext = container_path[container_path.rindex(".") + 1:]
source_file_name = f"{file_name}_{asset_name}_{path_id}.{ext}"
if source_file_name not in source_files:
continue
if match_pattern != None:
if re.match(match_pattern, source_file_name) == None:
continue
# Edit
change_flag = True
data = obj.read()
with open(os.path.join(source_path, source_file_name), "rb") as f:
source_data = f.read()
data.script = source_data
data.save()
# Important: Save the file as a new file(decompressed)
if change_flag:
with open(f"{output_path}/{file_name}_decompressed", "wb+") as f:
f.write(env.file.save())
inject_TextAsset("game/AssetBundles", "translated")