Intellij IDEA：无法提交文件：'utf8'编解码器无法解码位置9的字节0xcc

Question

IntelliJ IDEA 14.0.1

Plugin: jetbrains-bitbucket-connector

I'm trying to commit files, but get the error:

Error:transaction abort!

rollback completed abort:

decoding near 'C:\\Users\\ \\AppDa': 'utf8' codec can't decode byte 0xcc in position 9: invalid continuation byte!

Has anyone encountered this error? How can it be solved?

Thanks.

Answer 1

This probably isn't the answer which you're looking for but gives you some insight on what might be going on:

On most systems, file paths are made up from bytes since file systems were designed decades before Unicode. Unicode is retrofitted to them by interpreting the bytes as UTF-8 encoded strings. Unfortunately, there is no way to say "this is Cp-1251" and "this is UTF-8" inside of a file name. Therefore, the "convert file name to string" code relies on the platform's default encoding. NTFS solved the problem by always storing file names as Unicode (ignoring the local code page) but the names are translated into the local code page when you use a tool which displays them on screen.

And then comes Python 2 where Unicode was also retrofitted in a similar way. Python just has the advantage that you have two types of objects ( str and unicode ) so in theory, you can tell raw bytes and Unicode apart. The problems start when you get a bunch of bytes from somewhere and the logic says "this should be Unicode" - which happens when you read file names from disk.

In your case, the file system passes bytes which contain Cp1251 encoded characters to Python but the Python code tries to read them as UTF-8 encoded Unicode. For many characters (< code point 128), this works but it breaks for everything with a code point > 128. \\xCC is a common case here since UTF-8 uses this byte to encode all code points between 128 and 256. This is why you see this error so often in Europe - we use those characters a lot.

Now the people who created Mercurial are well aware of all this. Most of the time, Mercurial should just work. See https://www.mercurial-scm.org/pipermail/mercurial/2009-January/023762.html

As I see it, your problem could be caused by:

• Somehow, Windows used the local code page to create your home directory (unlikely)
• Mercurial gets the path as Unicode but for some reason, it thinks that the string is raw bytes and tried to decode using a UTF-8 decoder. Since the decoding is applied twice, this fails. Maybe you have an old version of Mercurial. Try to update.
• Maybe you showed us the wrong part of the error message and the problem is actually in a file which you tried to commit. In that case, we can ignore the odd characters in the error message. Make sure you use the correct encoding when you edit the file.

To see which one it is, I suggest to create a folder C:\\dev and work there. If this works, then there is something wrong with your home folder or Mercurial has a bug.

Answer 2

Error is saying that in your file location path there are few character which is not present in the utf-8 character set so the decoder is not able to decode the given file path and it is aborting the operation.

see the characters in the location path and correct it if there are any unknown character present in that

'C:\\Users\\ \\AppDa'

here the are showing that these characters are not able to decode by utf-8.

edit: Check your string with this tool to see in which encoding your character set is encoded. link to tool

then you can use that encoder but this is not a practical solution use utf-16 char set it is having large character set, and it vary by platform and language.

Answer 3

I had the same problem, if you use Mercurial also, then here is the solution:

• go to [project directory]/.hg
• open "hgrc" file
• below the [ui] insert username = my_name_only_utf_characters <mail@example.com>
• save & commit