info --apropos should decode/encode nodes/index entries

[Patrice Dumas]

Hello,

In general I think that info should decode nodes to UTF-8 or something
like that when they are in 8bit encodings and there is a need to search
in them, and, when outputting on the terminal, should encode to the
locale.  I attach an example with an index done in a latin1 encoded
file, the index entry with an accented letter is not found, hinting that
the latin1 encoded index entry is not decoded to UTF-8, and the index
entry encoded in latin1 is not well output.  The command line should
also be decoded to UTF-8.

My local is a mix of en_US.UTF-8 and fr_FR.UTF-8.

Here is an example showing that a non ascii word is not found in the
latin1 encoded info file, and the non ascii characters in the latin1
encoded entries are not output in the UTF-8 encoding of my locale:

$ tar xvf infodir.tar.gz -C /some/where/tmp/
$ export INFOPATH=/some/where/tmp/infodir
$ info -v infopath-no-defaults=On --apropos café
"(index_utf8)chap Index" -- UTF-8 café ≡ 这常 toto
$ info -v infopath-no-defaults=On --apropos toto
"(index_latin1)chap Index" -- latin1 � == � toto
"(index_utf8)chap Index" -- UTF-8 café ≡ 这常 toto

The index entries are:
* in latin1:
 latin1 é == ô toto

* in UTF-8 (ther are two chinese characters):
 UTF-8 café ≡ 这常 toto

-- 
Pat

infodir.tar.gz
Description: application/gzip