[Nmh-workers] mhshow not honoring charset spec

[Anders Eriksson]

I'm having a go at teaching nmh to show various charsets properly on 
my UTF-8 terminal using iconv. No luck, though. :-( It seems that the 
only mhshow-charset-xxx entries that are recognized are for 
xxx == iso-8859-1.

What I have is this:
showproc: mhshow
mhshow-show-text/plain: cat
mhshow-charset-iso-8859-1: %s | iconv -f iso-8859-1 -t utf-8 | less

It does the trick for an iso-8859-1 message. Changing the "-1" with "-9"
and testing with a an iso-8859-9 message makes it use the normal pager 
(less) on the body. My initial suspicion was that it somehow thought that
iso-8859-9 was the native locate, so I started to look into the code, 
and ended up looking at this in mhparse.c:

struct k2v Charset[] = {
    { "us-ascii",   CHARSET_USASCII },
    { "iso-8859-1", CHARSET_LATIN },
    { NULL,         CHARSET_UNKNOWN }  /* this one must be last! */
};

...which seems to be used to compare against the charset in the message:

    /* check if content specified a character set */
    if (*ap) {
        /* match character set or set to CHARSET_UNKNOWN */
        for (kv = Charset; kv->kv_key; kv++) {
            if (!mh_strcasecmp (*ep, kv->kv_key)) {
                   chset = *ep;
                    break;
            }
        }


...and this is used here:
     * If we can not handle character set natively,
     * then check profile for string to modify the
     * terminal or display method.
     *
     * termproc is for mhshow, though mhlist -debug prints it, too.
     */
     if (chset != NULL && !check_charset (chset, strlen (chset))) {
         snprintf (buffer, sizeof(buffer), "%s-charset-%s", invo_name, chset);
         if ((cp = context_find (buffer)))
             ct->c_termproc = getcpy (cp);
     }

I don't claim to understand all the details here, but it seems that
(and sprinkling printf's confirm) that chset is only set if a well-known
charset is used, and _then_ we go look for a mhshow-charset-xxx entry ??

I've seen no other reports on this on the net, but quite a few posts 
suggesting that any random charset can be handled using mhshow-charset-xxx
so I'm guessing I've gotten something backwards.

All pointers appreciated,
-Anders