How to find the character breaking the file encoding?

How to find the character breaking the file encoding?

[Marcin Borkowski]

Hello all,

I have a large UTF-8 file (about 1.5MB) to which I add stuff regularly.
Recently, Emacs started saving it as a binary file.  I suspect I somehow
inserted a non-UTF-8 sequence of bytes there.  Is there any way Emacs
can help me finding it (other than me manually bisecting the file until
I find the offending place)?

TIA,

-- 
Marcin Borkowski
http://mbork.pl

Re: How to find the character breaking the file encoding?

[Óscar Fuentes]

Marcin Borkowski <mbork@mbork.pl> writes:

> I have a large UTF-8 file (about 1.5MB) to which I add stuff regularly.
> Recently, Emacs started saving it as a binary file.  I suspect I somehow
> inserted a non-UTF-8 sequence of bytes there.  Is there any way Emacs
> can help me finding it (other than me manually bisecting the file until
> I find the offending place)?

Try using M-x encode-coding-region ENTERT utf-8 ENTER

With some luck this will mark the buffer as modified because it replaces
unencodable content with blanks (IIRC). Then you can use M-x
diff-buffer-with-file to see the line that contains the problematic
sequence.

Re: How to find the character breaking the file encoding?

[Stefan Monnier]

> inserted a non-UTF-8 sequence of bytes there.  Is there any way Emacs
> can help me finding it (other than me manually bisecting the file until
> I find the offending place)?

I think you can do:

    C-x RET r utf-8 RET

to force reading it as utf-8.  And then

    C-x C-s

should complain about the problematic element with a buffer that shows
you the offender and lets you click on it to jump to its location.


        Stefan

Re: How to find the character breaking the file encoding?

[Marcin Borkowski]

On 2020-02-05, at 04:09, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

>> inserted a non-UTF-8 sequence of bytes there.  Is there any way Emacs
>> can help me finding it (other than me manually bisecting the file until
>> I find the offending place)?
>
> I think you can do:
>
>     C-x RET r utf-8 RET
>
> to force reading it as utf-8.  And then
>
>     C-x C-s
>
> should complain about the problematic element with a buffer that shows
> you the offender and lets you click on it to jump to its location.

Thanks Stefan and Óscar.  Before you answered me, I killed the buffer
and visited the file again, and the problem disappeared.  Go figure...

Thanks,

-- 
Marcin Borkowski
http://mbork.pl