const char *
encoding_guess_parse_encoding (const char *encoding)
{
- const char *fallback;
-
if (encoding == NULL
|| !c_strcasecmp (encoding, "auto")
|| !c_strcasecmp (encoding, "auto,locale")
|| !c_strcasecmp (encoding, "locale"))
- fallback = locale_charset ();
+ return locale_charset ();
else if (!c_strncasecmp (encoding, "auto,", 5))
- fallback = encoding + 5;
+ return encoding + 5;
else
return encoding;
-
- return is_encoding_utf8 (fallback) ? "windows-1252" : fallback;
}
/* Returns true if ENCODING, which must be in one of the forms described at the
if (is_utf32 (data, n, get_le32))
return "UTF-32LE";
- if (!is_encoding_ascii_compatible (fallback_encoding)
- || !encoding_guess_tail_is_utf8 (data, n))
+ /* We've tried all the "giveaways" that make the encoding obvious. That
+ rules out, incidentally, all the encodings with multibyte units
+ (e.g. UTF-16, UTF-32). Our remaining goal is to try to distinguish UTF-8
+ from some ASCII-based fallback encoding. */
+
+ /* If the fallback encoding isn't ASCII compatible, give up. */
+ if (!is_encoding_ascii_compatible (fallback_encoding))
return fallback_encoding;
+ /* If the data we have clearly is not UTF-8, give up. */
+ if (!encoding_guess_tail_is_utf8 (data, n))
+ {
+ /* If the fallback encoding is UTF-8, fall back on something else.*/
+ if (is_encoding_utf8 (fallback_encoding))
+ return "windows-1252";
+
+ return fallback_encoding;
+ }
+
return "ASCII";
}
encoding_guess_tail_encoding (const char *encoding,
const void *data, size_t n)
{
- return (encoding_guess_tail_is_utf8 (data, n) != 0
- ? "UTF-8"
- : encoding_guess_parse_encoding (encoding));
+
+ if (encoding_guess_tail_is_utf8 (data, n) != 0)
+ return "UTF-8";
+ else
+ {
+ /* The data is not UTF-8. */
+ const char *fallback_encoding = encoding_guess_parse_encoding (encoding);
+
+ /* If the fallback encoding is UTF-8, fall back on something else.*/
+ if (is_encoding_utf8 (fallback_encoding))
+ return "windows-1252";
+
+ return fallback_encoding;
+ }
+
}
/* Returns an encoding guess based on ENCODING and the N bytes of text starting