CJKV Information Processing, 2nd Edition

Miscellaneous Algorithms

|

595

output[ixOut++] = input[ixIn++];

}

}

String output_string = new String(output);

return output_string.substring(0,ixOut);

}

}

Appendix C provides a half- to full-width katakana conversion program written in Perl.

It is dierent in that it is not based on Unicode, but rather supports EUC-JP and Shi-JIS

encodings.

Encoding Repair

ISO-2022-JP–encoded les oen become damaged or corrupt from soware that strips

out escape characters. Some programs have a tendency to lter out control characters

from les; the escape character (1 or 1), which is an essential part of ISO-

2022-JP encoding, is a control character. Luckily, there are ways to repair corrupted

ISO-2022-JP–encoded les.

You can make a few assumptions before you proceed to repair damaged ISO-2022-JP–

encoded les. e rst assumption is that the text stream begins, and also ends, in

one-byte mode. In addition, each line begins and ends in one-byte mode. e next as-

sumption is that the other characters that make up a complete escape sequence are still

intact. ese may include such strings as

$@, $B, (J, and (B. Depending on the n-byte-per-

character mode, you need to scan for dierent strings.

While in one-byte mode, you need only to scan for the string

$B or $@, which should

signify the beginning of two-byte mode. e chances of encountering such