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


Text::Kakasi - perl frontend to kakasi


  use Text::Kakasi;
  # functional
  $res = Text::Kakasi::getopt_argv('-JJ', '-c', '-w');
  $str = Text::Kakasi::do_kakasi($japanese_text);
  # object-oriented
  $obj = Text::Kakasi->new('-JJ', '-c', '-w');
  $str = $obj->get($japanese_text);


This module provides interface to kakasi (kanji kana simple inverter). kakasi is a set of programs and libraries which does what Japanese input methods do in reverse order. You feed Japanese and kakasi converts it to phonetic representation thereof. kakasi can also be used to tokenizing Japanese text. To find more about kakasi, see .

Text::Kakasi now features both functional and object-oriented APIs. functional APIs are 100% compatible with ver. 1.05. But to take advantage of "Perl 5.8 Features", you should use OOP APIs instead.

See Text::Kakasi::JP for the Japanese version of this document.

Functional APIs

Note Text::Kakasi:: is omitted. Text::Kakasi does not export these functions by default. You can import these function as follows;

  use Text::Kakasi qw/getopt_argv do_kakasi/;
$err = getopt_argv($arg1, $arg2, ...)

initializes kakasi with options options are the same as kakasi command. Here is the summery as of kakasi 2.3.4.

  -a[jE] -j[aE] -g[ajE] -k[ajKH]
  -E[aj] -K[ajkH] -H[ajkK] -J[ajkKH]
  -r{hepburn,kunrei} -p -s -f -c"chars" 
   [jisyo1, jisyo2,,,]

  Character Sets:
       a: ascii  j: jisroman  g: graphic  k: kana 
       (j,k     defined in jisx0201)
       E: kigou  K: katakana  H: hiragana J: kanji
       (E,K,H,J defined in jisx0208)

    -i: input coding system    -o: output coding system
    -r: romaji conversion system
    -p: list all readings (with -J option)
    -s: insert separate characters (with -J option)
    -f: furigana mode (with -J option)
    -c: skip chars within jukugo
        (with -J option: default TAB CR LF BLANK)
    -C: romaji Capitalize (with -Ja or -Jj option)
    -U: romaji Upcase     (with -Ja or -Jj option)
    -u: call fflush() after 1 character output
    -w: wakatigaki mode

Returns 0 on success and nonzero on failure.

Unlike version 1.x where you have to start the first argument with kakasi, you can omit that in version 2.x (adding kakasi does not harm so compatibility is preserved).

$result_str = do_kakasi($str)

apply kakasi to $str and returns result. If anything goes wrong it return undef.


closes dictionary files which are implicitly opened. This function is for backward compatibity only and you should never have to use this function today.

Object-Oriented APIs

As of 2.0, Text::Kakasi also offers OOP APIs.

$k = Text::Kakasi->new($args ...)

Constructs object. When argument is fed, it is the same as Text::Kakasi->new->set($args ...)

$k->set($args ...)

OOP interface to getopt_argv.

  my $k = Text::Kakasi->new;
  $k->set('-w'); # Text::Kakasi::getopt_argv('-w');

Unlike getopt_argv() which returns the status, set returns the object itself so you can go like this;

  my $tokenized = $k->set('-w')->get($raw_japanese);

To get the status of $k->set, use $k->error.

See also "Perl 5.8 Features".


returns the status of last method.

$result = $k->get($raw_japanese);

OOP interface to do_kakasi. The following codes are equivalent.

  # Functional
  getopt_argv('-w'); $result = do_kakasi($raw_japanese);
  # OOP

Perl 5.8 Features

Perl 5.8 introduces Encode module which transcodes various encodings. This module takes advantage of this feature but to keep backward compatibility with version 1.x, This feature is enabled only when you use OOP interface (version 1.x only provided functional APIs).

On Perl 5.8 and up, -iencoding and -oencodingare handled by Encode module so you can use encodings Kakasi does not suppport such as utf8. In other words,

  $result = $k->set(qw/-iutf8 -outf8 -w/)->get($utf8);

Is analogous to:

  $euc = encode('eucjp' => $utf8);
  $tmp = do_kakasi($euc);
  $result = decode('eucjp' => $tmp);

When you specify -outf8, $k->get will return the string with utf8 flag on.

You can suppress this feature by setting $Text::Kakasi::HAS_ENCODE to 0 in which case this feature is not used.




  (C) 1998, 1999, 2000 NOKUBI Takatsugu <>
  (C) 2003 Dan Kogai <>

There is no warranty for this free software. Anyone can modify and/or redistribute this module under GNU GENERAL PUBLIC LICENSE. See COPYING file that is included in the archive for more details.