Text::ANSI::Fold - Text folding library supporting ANSI terminal sequence and Asian wide characters with prohibition character handling.


Version 2.2102


    use Text::ANSI::Fold qw(ansi_fold);
    ($folded, $remain) = ansi_fold($text, $width, [ option ]);

    use Text::ANSI::Fold;
    my $f = Text::ANSI::Fold->new(width => 80, boundary => 'word');
    $f->configure(ambiguous => 'wide');
    ($folded, $remain) = $f->fold($text);

    use Text::ANSI::Fold;
    while (<>) {
        print join "\n",
            Text::ANSI::Fold->new(width => 40, text => $_)->chops;

    use Text::ANSI::Fold qw(:constants);
    my $fold = Text::ANSI::Fold->new(
        width     => 70,
        boundary  => 'word',
        linebreak => LINEBREAK_ALL,
        runin     => 4,
        runout    => 4,


Text::ANSI::Fold provides capability to divide a text into two parts by given width. Text can include ANSI terminal sequences and the width is calculated by its visible representation. If the text is divided in the middle of colored region, reset sequence is appended to the former text, and color recover sequence is inserted before the latter string.

This module also support Unicode Asian full-width and non-spacing combining characters properly. Japanese text formatting with head-or-end of line prohibition character is also supported. Set the linebreak mode to enable it.

Since the overhead of ANSI escape sequence handling is not significant when the data does not include them, this module can be used for normal text processing with small penalty.

Use exported ansi_fold function to fold original text, with number of visual columns you want to cut off the text.

    ($folded, $remain, $w) = ansi_fold($text, $width);

It returns a pair of strings; first one is folded text, and second is the rest.

Additional third result is visual width of the folded text. It is not always same as given width, and you may want to know how many columns returned string takes for further processing.

Negative width value is taken as unlimited. So the string is never folded, but you can use this to expand tabs and to get visual string width.

This function returns at least one character in any situation. If you provide Asian wide string and just one column as width, it trims off the first wide character even if it does not fit to given width.

Default parameter can be set by configure class method:

    Text::ANSI::Fold->configure(width => 80, padding => 1);

Then you don't have to pass second argument.

    ($folded, $remain) = ansi_fold($text);

Because second argument is always taken as width, use undef when using default width with additional parameter:

    ($folded, $remain) = ansi_fold($text, undef, padding => 1);

Some other easy-to-use interfaces are provided by sister module Text::ANSI::Fold::Util.


You can create an object to hold parameters, which is effective during object life time. For example,

    my $f = Text::ANSI::Fold->new(
        width => 80,
        boundary => 'word',

makes an object folding on word boundaries with 80 columns width. Then you can use this without parameters.


Use configure method to update parameters:

    $f->configure(padding => 1);

Additional parameter can be specified on each call, and they precede saved value.

    $f->fold($text, width => 40);


A fold object can hold string inside by text method.


And folded string can be taken by retrieve method. It returns empty string if nothing remained.

    while ((my $folded = $f->retrieve) ne '') {
        print $folded;
        print "\n" if $folded !~ /\n\z/;

Method chops returns chopped string list. Because the text method returns the object itself when called with a parameter, you can use text and chops in series:

    print join "\n", $f->text($string)->chops;

Actually, text can be set by new or configure method through text parameter. Next program just works.

    use Text::ANSI::Fold;
    while (<>) {
        print join "\n",
            Text::ANSI::Fold->new(width => 40, text => $_)->chops;

When using chops method, width parameter can take array reference, and chops text into given width list.

    my $fold = Text::ANSI::Fold->new;
    my @list = $fold->text("1223334444")->chops(width => [ 1, 2, 3 ]);
    # return ("1", "22", "333") and keep "4444"

If the width value is 0, it returns empty string. Negative width value takes all the rest of stored string.

Method text has an lvalue attribute, so it can be assigned to, as well as can be a subject of mutating operator such as s///.


Option parameter can be specified as name-value list for ansi_fold function as well as new and configure method.

    ansi_fold($text, $width, boundary => 'word', ...);

    Text::ANSI::Fold->configure(boundary => 'word');

    my $f = Text::ANSI::Fold->new(boundary => 'word');

    $f->configure(boundary => 'word');
width => n, [ n, m, ... ]

Specify folding width. Negative value means all the rest.

Array reference can be specified but works only with chops method, and retunrs empty string for zero width.

boundary => word or space

Option boundary takes word and space as a valid value. These prohibit to fold a line in the middle of ASCII/Latin sequence. Value word means a sequence of alpha-numeric characters, and space means simply non-space printables.

This operation takes place only when enough space will be provided to hold the word on next call with same width.

If the color of text is altered within a word, that position is also taken as an boundary.

padding => bool

If padding option is given with true value, margin space is filled up with space character. Default is 0. Next code fills spaces if the given text is shorter than 80.

    ansi_fold($text, 80, padding => 1);

If an ANSI Erase Line sequence is found in the string, color status at the position is remembered, and padding string is produced in that color.

padchar => char

padchar option specifies character used to fill up the remainder of given width.

    ansi_fold($text, 80, padding => 1, padchar => '_');
prefix => string | coderef

prefix string is inserted before remained string if it is not empty. This is convenient to produce indented series of text by chops interface.

If the value is reference to subroutine, its result is used as a prefix string.

ambiguous => narrow or wide

Tells how to treat Unicode East Asian ambiguous characters. Default is narrow which means single column. Set wide to tell the module to treat them as wide character.

discard => [ EL, OSC ]

Specify the list reference of control sequence name to be discarded. EL means Erase Line; OSC means Operating System Command, defined in ECMA-48. Erase Line right after RESET sequence is always kept.

linebreak => mode
runin => width
runout => width

These options specify the behavior of line break handling for Asian multi byte characters. Only Japanese is supported currently.

If the cut-off text start with space or prohibited characters (e.g. closing parenthesis), they are ran-in at the end of current line as much as possible.

If the trimmed text end with prohibited characters (e.g. opening parenthesis), they are ran-out to the head of next line, if it fits to maximum width.

Default linebreak mode is LINEBREAK_NONE and can be set one of those:


Import-tag :constants can be used to access these constants.

Option runin and runout is used to set maximum width of moving characters. Default values are both 2.

expand => bool
tabstop => n
tabhead => char
tabspace => char

Enable tab character expansion.

Default tabstop is 8 and can be set by tabstop option.

Tab character is converted to tabhead and following tabspace characters. Both are white space by default.

tabstyle => style

Set tab expansion style. This parameter set both tabhead and tabspace at once according to the given style name. Each style has two values for tabhead and tabspace.

If two style names are combined, like symbol,space, use symbols's tabhead and space's tabspace.

Currently these names are available.

                "\N{SYMBOL FOR SPACE}" ],
    shade  => [ "\N{MEDIUM SHADE}",
                "\N{LIGHT SHADE}" ],
    block  => [ "\N{LOWER ONE QUARTER BLOCK}",
                "\N{LOWER ONE EIGHTH BLOCK}" ],
    bar    => [ "\N{BOX DRAWINGS HEAVY RIGHT}",
                "\N{BOX DRAWINGS LIGHT HORIZONTAL}" ],
    dash   => [ "\N{BOX DRAWINGS HEAVY RIGHT}",

Below are styles providing same character for both tabhead and tabspace.

    dot          => '.',
    space        => ' ',
    emspace      => "\N{EM SPACE}",
    middle-dot   => "\N{MIDDLE DOT}",
    arrow        => "\N{RIGHTWARDS ARROW}",
    double-arrow => "\N{RIGHTWARDS DOUBLE ARROW}",
    triple-arrow => "\N{RIGHTWARDS TRIPLE ARROW}",
    white-arrow  => "\N{RIGHTWARDS WHITE ARROW}",
    wave-arrow   => "\N{RIGHTWARDS WAVE ARROW}",
    squat-arrow  => "\N{SQUAT BLACK RIGHTWARDS ARROW}",
    squiggle     => "\N{RIGHTWARDS SQUIGGLE ARROW}",
    cuneiform    => "\N{CUNEIFORM SIGN TAB}",


Next code implements almost fully-equipped fold command for multi byte text with Japanese prohibited character handling.

    #!/usr/bin/env perl
    use strict;
    use warnings;
    use open IO => 'utf8', ':std';
    use Text::ANSI::Fold qw(:constants);
    my $fold = Text::ANSI::Fold->new(
        width     => 70,
        boundary  => 'word',
        linebreak => LINEBREAK_ALL,
        runin     => 4,
        runout    => 4,
    $, = "\n";
    while (<>) {
        print $fold->text($_)->chops;


Collection of ANSI related tools.


Distribution and repository.


Command line utility using Text::ANSI::Fold.


Collection of utilities using Text::ANSI::Fold module.


Text::Tabs compatible tab expand/unexpand module using Text::ANSI::Fold as a backend processor.


Text::ANSI::Fold was originally implemented in sdif command for long time, which provide side-by-side view for diff output. It is necessary to process output from cdif command which highlight diff output using ANSI escape sequences.

Text::ANSI::Util, Text::ANSI::WideUtil

These modules provide a rich set of functions to handle string contains ANSI color terminal sequences. In contrast, Text::ANSI::Fold provides simple folding mechanism with minimum overhead. Also sdif need to process other than SGR (Select Graphic Rendition) color sequence, and non-spacing combining characters, those are not supported by these modules.

ANSI escape code definition.

Requirements for Japanese Text Layout, W3C Working Group Note 11 August 2020


ECMA-48: Control Functions for Coded Character Sets


Kazumasa Utashiro


Copyright 2018-2023 Kazumasa Utashiro.

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