File Coverage

blib/lib/HTML/Strip/Whitespace.pm

Criterion	Covered	Total	%
statement	89	98	90.8
branch	19	26	73.0
condition	9	10	90.0
subroutine	21	23	91.3
pod	1	1	100.0
total	139	158	87.9

package HTML::Strip::Whitespace;
$HTML::Strip::Whitespace::VERSION = '0.2.0';
use strict;
use warnings;

use 5.016;

package HTML::Strip::Whitespace::State;
$HTML::Strip::Whitespace::State::VERSION = '0.2.0';
sub new
{
    my $class = shift;
    my $self  = {};
    bless $self, $class;
    $self->initialize(@_);
    return $self;
}

sub to_array
{
    my $v = shift;
    return ( ref($v) eq "ARRAY" ? (@$v) : $v );
}

sub initialize
{
    my $self = shift;
    my %args = (@_);
    $self->{'prev'} = undef;
    $self->{'next'} = undef;
    $self->{'this'} = undef;
    $self->{'parser'} =
        HTML::TokeParser::Simple->new( to_array( $args{'parser_args'} ) );

    $self->{'strip_newlines'} = $args{'strip_newlines'} || 0;
    $self->{'out_fh'} = $args{'out_fh'};

    # Get the first element to initialize the parser
    # Otherwise the first call to next_state would return undef;
    $self->next_state();

    return 0;
}

sub next_state
{
    my $self = shift;
    ( $self->{'prev'}, $self->{'this'}, $self->{'next'} ) =
        ( $self->{'this'}, $self->{'next'}, $self->{'parser'}->get_token() );
    if ( !defined( $self->{'this'} ) )
    {
        return undef;
    }
    return 1;
}

sub prev
{
    my $self = shift;
    return $self->{'prev'};
}

sub next
{
    my $self = shift;
    return $self->{'next'};
}

sub this
{
    my $self = shift;
    return $self->{'this'};
}

sub text_strip
{
    my $self = shift;

    # my $p = $self->prev();
    # my $n = $self->next();

    my $text = $self->this()->as_is();

    $text =~ s{([\s\n]+)}{($1 =~ /\n/) ? "\n" : " "}eg;

    return $text;
}

my %preserving_start_tags = ( 'pre' => 1, );

sub is_preserving_start_tag
{
    my $self = shift;
    my $t    = $self->this();
    if ( $t->is_start_tag()
        && exists( $preserving_start_tags{ $t->get_tag() } ) )
    {
        return $t->get_tag();
    }
    return undef;
}

sub handle_text
{
    my $state = shift;

    if ( $state->this->is_text() )
    {
        $state->out( $state->text_strip() );
        return 0;
    }
    else
    {
        return 1;
    }
}

sub out
{
    my $self   = shift;
    my $what   = shift;
    my $out_fh = $self->{'out_fh'};

    if ( ref($out_fh) eq "CODE" )
    {
        &{$out_fh}($what);
    }
    elsif ( ref($out_fh) eq "SCALAR" )
    {
        $$out_fh .= $what;
    }
    elsif ( ref($out_fh) eq "GLOB" )
    {
        print { *{$out_fh} } $what;
    }

    return 0;
}

sub out_this
{
    my $state = shift;
    $state->out( $state->this()->as_is() );
}

sub process
{
    my $state = shift;

    my $tag_type;

    while ( $state->next_state() )
    {
        if ( !$state->handle_text() )
        {
            # Text was handled
        }

        # If it's a preserving start tag, preserve all the text inside it.
        # This is for example, a  tag in which the spaces matter.
        elsif ( $tag_type = $state->is_preserving_start_tag() )
        {
            my $do_once = 1;
            while ( $do_once || $state->next_state() )
            {
                $do_once = 0;
                $state->out_this();
                last if ( $state->this()->is_end_tag($tag_type) );
            }
        }
        else
        {
            $state->out_this();
        }
    }

    # Return 0 on success.
    return 0;
}

package HTML::Strip::Whitespace;

use 5.008;
use strict;
use warnings;

use HTML::TokeParser::Simple;

use parent 'Exporter';
use vars qw(@EXPORT @EXPORT_OK %EXPORT_TAGS);

%EXPORT_TAGS = (
    'all' => [
        qw(
            html_strip_whitespace
            )
    ]
);

@EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} } );

@EXPORT = qw(

);

# Preloaded methods go here.

sub html_strip_whitespace
{
    my %args   = (@_);
    my $source = $args{'source'}
        or die "source argument not specified.";
    my $strip_newlines = $args{'strip_newlines'} || 0;
    my $out_fh = $args{'out'}
        or die "out argument not specified.";
    my $state = HTML::Strip::Whitespace::State->new(
        'parser_args'    => $source,
        'strip_newlines' => $strip_newlines,
        'out_fh'         => $out_fh,
    );

    return $state->process();
}

# Autoload methods go after =cut, and are processed by the autosplit program.

1;

__END__