NAME¶

kcc - Kanji code coverter with encoding auto detection

SYNOPSIS¶

kcc [ -IOchnvxz ] [ -b bufsize ] [ file ] ...

DESCRIPTION¶

kcc is a filter that reads file sequencially, converts kanji encodings and output to stdout. If no file is specified, or specified - as filename, it read from stdin. You can specify kanji encodings for input/output. However, kcc detect input encodig automatically, if you don't specify input encoding. Available kanji encodings are JIS (7 bit and/or 8 bit), Shift JISEUCDEC. For input encoding, you can mix when these are pair of one of EUC DEC or Shift JIS and 7 bit JIS. SI/SOESC(I are recognized as halfwidth of JIS.

OPTIONS¶

-O

-IO

I for input kanji encoding¡¤O for output kanji encoding. When no input encoding specified, it will be detected automatically, and if both of input/output aren't specified, output encoding is 7 bit JIS.

You can specify one of the followings for the input encoding option, I.

You can specify one of the followings for output encoding option, O.

By XY in O option, You can specify which escape sequence used in JIS encoding. BJ is default. Supplimental kanji designation is fixed to ESC$(D

-v

outputs result of input encoding detection to stderr.

-x

Extension mode. By auto detection of input encodings, recognize user-defined characters and extended character region ( out of range of EUC, undefined halfwidth kana, control character, C1 area and/or extended character region Shift C1 JIS ). Distinguish between DEC and EUC is done in this mode.

-z

Shrink mode. Don't recognize halfwidth kana (except 7 bit JIS ) with input encoding detection. With this option, accuracy of auto detection of input encodings becomes much better for file without halfwidth kana.

-h

Normally, When converted halfwidth kana to DEC , it becomes fullwidth Katakana. With this option, it becomes Hiragana.

-n

user-defined characters, extended characters and supplimental kanji characters areconverted to fullwidth white box, and undefined region of halfwidth kana are converted to halfwidth centered dot.

-b bufsize

specify buffer size. 8kbytes is default.

-c

don't convert but check input encoding and print result to stdout. Different with normal auto-detection, whole contents of file is checked. However, when inconsistency of encodings is found, abort reading and print "data". Options except -x¡¤-z are ignored.

EXAMPLES¶

% kcc -e file

Input encoding are detect automatically, and output is in EUC encoding.

% kcc -sj file1 file2

Two files in Shift JIS concatinated with converting to JIS.

% command | kcc -k+J

output of command are converted to JIS(JIS JIS X0208 JIS JIS Roman¡¤ ESC(I Halfwidth Kana JIS )

% kcc -c file

Encoding of contents of file is detected(no conversion)

BUG¶

Auto detection of input encoding is well done for normal case, however, it has the following problems. 7 bit JIS is recognized by escape sequence in certain. EUC and DEC are the same (refered as EUC series). Halfwidth kana of 8 bit JIS is the same as halfwidth kana of Shift JIS (refered as Shift JIS series). However, EUC series and JIS , which are both 8 bit encoding, are sharing the same regions widely. So, the problem in auto detection is detection of these 2 encodings. Detection of EUC series/Shift JIS series is done in line by line, When it is found that it's not Shift JIS series, or it's not EUC series, encoding is determined. When inconsistensy found, it will be treated as "data" and contents of output is not guaranteed. While determined between EUC series/Shift JIS series after 8bit code found, conversions are pending and put input data in buffer, however, buffer is fulled, it assumes it's EUC series and forces to start conversion. Rationale. Usually, we can assume that documents with kanji include JIS non-kanji or JIS first standard, it can be detected in certain if it is Shift JIS , which does not share region with EUC. So if it can't be determined, it's very likely to be EUC. 8 bit JIS and it has always even number of halfwidth kana sequences, then it will be wrongly detected as EUC kanji. Be ceraful. If input encoding doesn't have halfwidth kana, use -z and accuracy of detection become much better. This is because shared region are restricted to area of JIS second standards. Extended region of Shift JIS user-defined area of EUC, control characters C1 of EUC, undefined region of halfwidth kana of EUC are out of range of auto detection, so it will fails to detect encodings if input has these characters. Use -x option to specify extended mode, or specify input code.

SEE ALSO¶

cat(1)

NOTES¶

Usually, user-defined characters, extended characters, supplimental kanji characters are mapped respectively. However characters that is out of range of extended characters become FCFC in hexadecimal when converted to Shift JIS. Although control character region C1 of EUC and DEC remains when converted to JIS , these will be deleted when converted to Shift JIS Undefined area of halfwidth kana become halfwidth centered dot when convered to Shift JIS Halfwidth kana become fullwidth kana when converted to DEC. When output is JIS encoding, control characters such as newline, TAB, DEL and white space (halfwidth) will be output in ASCII mode. When encoding of input is detected wrongly, or input undefined character for expected character sets, output is indefined. This manual are translated by Fumitoshi UKAI <ukai@debian.or.jp> for Debian system, but you can use it for any purpose.