|
There are two possibilities that I can think of, apart from you being mistaken.
(i) The calibre conversion might be doing something wrong with non-ascii characters, either interpreting utf8 as Window Latin-1 or vice versa
(ii) The calibre decoding of the mobi might be getting the trailing characters in each segment of the mobipocket file wrong.
But it seems unlikely to me that either thing would be wrong, so I'm guessing it must be a third problem I can't think of.
You could try extracting the contents of the mobipocket file with Mobiunpack to get and HTML file, add that to calibre, and see if that will convert cleanly.
Also, checking to see whether there are errors in the mobipocket file when viewed with the calibre file viewer would help pin down the problem.
Best would be to find a DRM-free sample file that shows this problem, so that the calibre developers could see the problem for themselves.
|