ISO charsets: lookup_charset

	The enumerated type [[ISO_charset]] contains the list of all
	known charsets. The function [[lookup_charset]] tries to map a
	string to one of them. The comparison is done case
	insensitively and while ignoring everything that is not a
	letter or digit. Therefore, [["ISO-8859_1"]] and
	[["ISO8859.1"]] and [["ISO.8859/1"]] will all be mapped to
	[[ISO8859_1]].

<<*>>=
#include <config.h>
#include <str.h>

EXPORT typedef enum {
    US_ASCII, ISO8859_1, ISO8859_2, ISO8859_3, ISO8859_4, ISO8859_5,
    ISO8859_6, ISO8859_7, ISO8859_8, ISO8859_9,
} ISO_charset;

EXPORT char *ISO_charset_repr[] = {
    "US-ASCII", "ISO8859-1", "ISO8859-2", "ISO8859-3", "ISO8859-4",
    "ISO8859-5", "ISO8859-6", "ISO8859-7", "ISO8859-8", "ISO8859-9",
};

#define charset2str(c) (ISO_charset_repr[c])
EXPORTDEF(charset2str(c))

EXPORT ISO_charset str2charset(char *s)
{
    char t[256];
    int i;

    for (i = 0; *s && i < sizeof(t) - 1; s++)
	if (isalnum(*s)) t[i++] = *s;
    t[i] = '\0';

    if (case_eq(t, "USASCII")) return US_ASCII;
    if (case_eq(t, "ISO88591")) return ISO8859_1;
    if (case_eq(t, "ISO88592")) return ISO8859_2;
    if (case_eq(t, "ISO88593")) return ISO8859_3;
    if (case_eq(t, "ISO88594")) return ISO8859_4;
    if (case_eq(t, "ISO88595")) return ISO8859_5;
    if (case_eq(t, "ISO88596")) return ISO8859_6;
    if (case_eq(t, "ISO88597")) return ISO8859_7;
    if (case_eq(t, "ISO88598")) return ISO8859_8;
    if (case_eq(t, "ISO88599")) return ISO8859_9;
    return US_ASCII;
}