The London Perl and Raku Workshop takes place on 26th Oct 2024. If your company depends on Perl, please consider sponsoring and/or attending.


Jcode - Japanese Charset Handler


 use Jcode;
 # traditional
 Jcode::convert(\$str, $ocode, $icode, "z");
 # or OOP!
 print Jcode->new($str)->h2z->tr($from, $to)->utf8;


<Japanese document is now available as Jcode::Nihongo. > supports both object and traditional approach. With object approach, you can go like;

  $iso_2022_jp = Jcode->new($str)->h2z->jis;

Which is more elegant than:

  $iso_2022_jp = $str;
  &jcode::convert(\$iso_2022_jp, 'jis', &jcode::getcode(\$str), "z");

For those unfamiliar with objects, still supports getcode() and convert().

If the perl version is 5.8.1, Jcode acts as a wrapper to Encode, the standard charset handler module for Perl 5.8 or later.


Methods mentioned here all return Jcode object unless otherwise mentioned.


$j = Jcode->new($str [, $icode])

Creates Jcode object $j from $str. Input code is automatically checked unless you explicitly set $icode. For available charset, see getcode below.

For perl 5.8.1 or better, $icode can be any encoding name that Encode understands.

  $j = Jcode->new($european, 'iso-latin1');

When the object is stringified, it returns the EUC-converted string so you can <print $j> instead of <print $j->euc>.

Passing Reference

Instead of scalar value, You can use reference as


This saves time a little bit. In exchange of the value of $str being converted. (In a way, $str is now "tied" to jcode object).

$j->set($str [, $icode])

Sets $j's internal string to $str. Handy when you use Jcode object repeatedly (saves time and memory to create object).

 # converts mailbox to SJIS format
 my $jconv = new Jcode;
 $/ = 00;
     print $jconv->set(\$_)->mime_decode->sjis;
$j->append($str [, $icode]);

Appends $str to $j's internal string.

$j = jcode($str [, $icode]);

shortcut for Jcode->new() so you can go like;

Encoded Strings

In general, you can retrieve encoded string as $j->encoded.

$sjis = jcode($str)->sjis
$euc = $j->euc
$jis = $j->jis
$sjis = $j->sjis
$ucs2 = $j->ucs2
$utf8 = $j->utf8

What you code is what you get :)

$iso_2022_jp = $j->iso_2022_jp

Same as $j->h2z->jis. Hankaku Kanas are forcibly converted to Zenkaku.

For perl 5.8.1 and better, you can also use any encoding names and aliases that Encode supports. For example:

  $european = $j->iso_latin1; # replace '-' with '_' for names.

FYI: Encode::Encoder uses similar trick.


For perl is 5.8.1 or better, Jcode stores the internal string in UTF-8. Any character that does not map to ->encoding are replaced with a '?', which is Encode standard.

  my $unistr = "\x{262f}"; # YIN YANG
  my $j = jcode($unistr);  # $j->euc is '?'

You can change this behavior by specifying fallback like Encode. Values are the same as Encode. Jcode::FB_PERLQQ, Jcode::FB_XMLCREF, Jcode::FB_HTMLCREF are aliased to those of Encode for convenice.

  print $j->fallback(Jcode::FB_PERLQQ)->euc;   # '\x{262f}'
  print $j->fallback(Jcode::FB_XMLCREF)->euc;  # '&#x262f;'
  print $j->fallback(Jcode::FB_HTMLCREF)->euc; # '&#9775;'

The global variable $Jcode::FALLBACK stores the default fallback so you can override that by assigning the value.

  $Jcode::FALLBACK = Jcode::FB_PERLQQ; # set default fallback scheme
[@lines =] $jcode->jfold([$width, $newline_str, $kref])

folds lines in jcode string every $width (default: 72) where $width is the number of "halfwidth" character. Fullwidth Characters are counted as two.

with a newline string spefied by $newline_str (default: "\n").

Rudimentary kinsoku suppport is now available for Perl 5.8.1 and better.

$length = $jcode->jlength();

returns character length properly, rather than byte length.

Methods that use MIME::Base64

To use methods below, you need MIME::Base64. To install, simply

   perl -MCPAN -e 'CPAN::Shell->install("MIME::Base64")'

If your perl is 5.6 or better, there is no need since MIME::Base64 is bundled.

$mime_header = $j->mime_encode([$lf, $bpl])

Converts $str to MIME-Header documented in RFC1522. When $lf is specified, it uses $lf to fold line (default: \n). When $bpl is specified, it uses $bpl for the number of bytes (default: 76; this number must be smaller than 76).

For Perl 5.8.1 or better, you can also encode MIME Header as:

  $mime_header = $j->MIME_Header;

In which case the resulting $mime_header is MIME-B-encoded UTF-8 whereas $j->mime_encode() returnes MIME-B-encoded ISO-2022-JP. Most modern MUAs support both.


Decodes MIME-Header in Jcode object. For perl 5.8.1 or better, you can also do the same as:

  Jcode->new($str, 'MIME-Header')

Hankaku vs. Zenkaku


Converts X201 kana (Hankaku) to X208 kana (Zenkaku). When $keep_dakuten is set, it leaves dakuten as is (That is, "ka + dakuten" is left as is instead of being converted to "ga")

You can retrieve the number of matches via $j->nmatch;


Converts X208 kana (Zenkaku) to X201 kana (Hankaku).

You can retrieve the number of matches via $j->nmatch;

Regexp emulators

To use ->m() and ->s(), you need perl 5.8.1 or better.

$j->tr($from, $to, $opt);

Applies tr/$from/$to/ on Jcode object where $from and $to are EUC-JP strings. On perl 5.8.1 or better, $from and $to can also be flagged UTF-8 strings.

If $opt is set, tr/$from/$to/$opt is applied. $opt must be 'c', 'd' or the combination thereof.

You can retrieve the number of matches via $j->nmatch;

The following methods are available only for perl 5.8.1 or better.

$j->s($patter, $replace, $opt);

Applies s/$pattern/$replace/$opt. $pattern and replace must be in EUC-JP or flagged UTF-8. $opt are the same as regexp options. See perlre for regexp options.

Like $j->tr(), $j->s() returns the object itself so you can nest the operation as follows;

  $j->tr("a-z", "A-Z")->s("foo", "bar");
[@match = ] $j->m($pattern, $opt);

Applies m/$patter/$opt. Note that this method DOES NOT RETURN AN OBJECT so you can't chain the method like $j->s().

Instance Variables

If you need to access instance variables of Jcode object, use access methods below instead of directly accessing them (That's what OOP is all about)

FYI, Jcode uses a ref to array instead of ref to hash (common way) to optimize speed (Actually you don't have to know as long as you use access methods instead; Once again, that's OOP)


Reference to the EUC-coded String.


Input charcode in recent operation.


Number of matches (Used in $j->tr, etc.)


($code, [$nmatch]) = getcode($str)

Returns char code of $str. Return codes are as follows

 ascii   Ascii (Contains no Japanese Code)
 binary  Binary (Not Text File)
 euc     EUC-JP
 sjis    SHIFT_JIS
 jis     JIS (ISO-2022-JP)
 ucs2    UCS2 (Raw Unicode)
 utf8    UTF8

When array context is used instead of scaler, it also returns how many character codes are found. As mentioned above, $str can be \$str instead. Users: This function is 100% upper-conpatible with jcode::getcode() -- well, almost;

 * When its return value is an array, the order is the opposite;
   jcode::getcode() returns $nmatch first.

 * jcode::getcode() returns 'undef' when the number of EUC characters
   is equal to that of SJIS.  Jcode::getcode() returns EUC.  for there is no in-betweens. 
Jcode::convert($str, [$ocode, $icode, $opt])

Converts $str to char code specified by $ocode. When $icode is specified also, it assumes $icode for input string instead of the one checked by getcode(). As mentioned above, $str can be \$str instead. Users: This function is 100% upper-conpatible with jcode::convert() !


For perl is 5.8.1 or later, Jcode acts as a wrapper to Encode. Meaning Jcode is subject to bugs therein.


This package owes a lot in motivation, design, and code, to the for Perl4 by Kazumasa Utashiro <>.

Hiroki Ohzaki <> has helped me polish regexp from the very first stage of development.

JEncode by has inspired me to integrate Encode to Jcode. He has also contributed Japanese POD.

And folks at Jcode Mailing list <>. Without them, I couldn't have coded this far.





Copyright 1999-2005 Dan Kogai <>

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.