Re: [tlug] Reading kanji file name from Mac OS X

[David Riggs <dariggs@example.com>]

Thanks for the suggestions Stephen, I can sometimes read some of the 
file names in emacs, but its a real hassel.

It seems that using WinXP is indeed the best way to handle this-- the 
Mac OS kanji file names are a mess.

Just for the record, here is my little note to myself about how to bring 
over a dir tree of Mac files with kanji and spaces and upper case ascii 
in the file names and translate from their names in sjis to utf-8, also 
converting spaces to "." and lowercasing ascii. And convert content too!

Copy to usb in WinXP, mount it in Linux as -o iocharset=sjis and then do
the following to convert to utf-8 and no-blank file names.

Go to the top level dir that contains the files from Mac. To clean up
Mac _directory_ names of blanks, converting the blank to a "."  run 
repeatedly until no errors are reported:

find * -type d -print0 | xargs -0 rename 's/ /./g'

(Several runs may be needed due to the problem of renaming a folder
that has already had a higher level rename done because it is itself
contained in folder that has a blank in the name. To properly separate 
file names despite the blanks,  -print0 directsl to use
null instead of blank for separators, and xargs -0 recognizes that.)

Then run for _files_ (should not be any more errors):

find * -type f -print0 | xargs -0 rename 's/ /./g'

Then convert kanji file names to utf-8, and also lowercase ascii. If
there are garbage, or non-sjis kanji names, "convmv" will report it and 
stop and you may have to correct non-sjis names manually, then rerun.
"convmv" does its own recursion into directories.

convmv -r -f sjis -t utf-8 --notest --lower *

Then convert data, both line breaks and encoding, (should be no blanks 
in names, but just in case use -printf0):

find * -type f -print0 |xargs -0 recode -f sjis/cl..utf-8/


Hope it is useful to someone,

David Riggs

Kyoto