stevenlele

> as Simplified Chinese variant font usually doesn't take precedence over traditional or Japanese font Isn't setting `` sufficient? I think modern browsers should be able to handle font family...

经过分析和查找，在网上找到了编码的算法，来源是分析他们浏览器的《[某度base64魔改算法还原](https://mp.weixin.qq.com/s/Ejf5QTO2dd0LaZqnvcIKXg)》。 文章里的分析和代码太乱了，用 Python 重写了一下，在这里贴一下解码的部分。Python 3.8 以上调用 `decode_bin()` 即可导出 UTF-16LE 的文本格式。 ```python MASK = 0x2D382324 TABLE = b"qogjOuCRNkfil5p4SQ3LAmxGKZTdesvB6z_YPahMI9t80rJyHW1DEwFbc7nUVX2-" DECODE_TABLE = bytes.maketrans(TABLE, bytes(range(64))) def decode(data: bytes): assert len(data) % 4 == 2...

> 很棒，您有兴趣提 pr 吗 我没怎么写过 C# 也没有环境，对项目也不熟悉，如果你方便的话就转写一下吧。上面有别人上传的测试的文件，如果需要的话我也可以提供解码后的格式样本。 如果可以的话麻烦在提交信息里面加上一行： ``` Co-authored-by: stevenlele ```

https://github.com/studyzy/imewlconverter/blob/3012c564f282da6fafed721cd27599397013c244/src/ImeWlConverterCore/IME/BaiduPinyinBackup.cs#L31-L32 程序确实是按照 `` → `` → `` 的顺序导出的，应该不会出现乱序。

然后英文词的格式是： ``` API 0 0 4 1707932168 ``` 分隔符是 `\t`，词频目测是 `[3]`，至于 `[1]` 和 `[2]` 的意义不明确。我这里 `[1]` 有少量的 1、2、5 等值，`[2]` 都是 0。

This guy is making an attempt to add a UI toggle for #3587 (that is, allow typing letters to match Chinese characters) although his approach is wrong. Pinyin support was...

As I mentioned, the current algorithm has severe performance issue, see https://github.com/microsoft/PowerToys/issues/3587#issuecomment-1382859784, so it wouldn't be ideal to release such an experimental feature.