发表于 2018-12-24   |   标签 Python   |  

由于excite关闭了网页翻译功能，google的网页翻译有时候也不太灵光，就去看了下jbeijing的翻译API调用，正好VNR中已有相关代码，就学习一下。

def transferUnicode(self, text, to, fr=CODEPAGE_JA):  
    """
    @param  text  unicode  not str
    @param  to  uint  code page
    @param  fr  uint  code page
    @raise  WindowsError, AttributeError
    From IDA:
    int __cdecl JC_Transfer_Unicode(int, UINT CodePage, int, int, int, LPCWSTR lpWideCharStr, int, int, int, int)
    Values are get by runtime debugging.

    Guessed
    int __cdecl JC_Transfer_Unicode(
      HWND hwnd,
      UINT fromCodePage,
      UINT toCodePage,
      int unknown,
      int unknown,
      LPCWSTR from,
      LPWSTR to,
      int &toCapacity,
      LPWSTR buffer,
      int &bufferCapacity)

    Note: This function is not thread-safe!
    Using persistent buffers is faster, but not thread-safe!
    """
    size = BUFFER_SIZE
    if self.BUFFER_THREAD_SAFE:
      # Limit buffer size would result in crash ... no idea why
      #size = min(size, len(text) * 10) # the translation should be no more larger than 10 times of Japanese
      out = ctypes.create_unicode_buffer(size)
      buf = ctypes.create_unicode_buffer(size)
    else:
      out = self.buffer1
      buf = self.buffer2
    outsz = ctypes.c_int(size)
    bufsz = ctypes.c_int(size)
    #self.dll.JC_Transfer_Unicode(
    self.JC_Transfer_Unicode(
      0,    # int, unknown
      fr,   # uint from, supposed to be 0x3a4 (cp932)
      to,   # uint to, eighter cp950 or cp932
      1,    # int, unknown
      1,    # int, unknown
      text,
      out,  # wchar_t*
      ctypes.byref(outsz), # &int
      buf,  # wchar_t*
      ctypes.byref(bufsz)) # &int
    return out.value

这段代码是调用JBJCT.dll中的方法(看注释大神是用IDA反编译dll在看函数的参数)，功能是将unicode文本从日文翻译成中文，下面是运行效果

实际运用中发现一个问题，翻译的文本是有上限的，如果原文是太长的文章，返回的译文可能会缺失（大概5k以上不行）