view release on metacpan or  search on metacpan

lib/Aozora2Epub.pm  view on Meta::CPAN

package Aozora2Epub;
use utf8;
use strict;
use warnings;
use Aozora2Epub::Gensym;
use Aozora2Epub::CachedGet qw/http_get/;
use Aozora2Epub::Epub;
use Aozora2Epub::XHTML;
use Path::Tiny;
use URI;
use HTML::Escape qw/escape_html/;

use base qw(Class::Accessor);
__PACKAGE__->mk_accessors(qw/files title author epub bib_info notation_notes/);

our $VERSION = '0.05';

our $AOZORA_GAIJI_URL = 'https://www.aozora.gr.jp/gaiji/';
our $AOZORA_CARDS_URL = 'https://www.aozora.gr.jp/cards';

sub _base_url {
    my $base = shift;
    $base =~ s{[^/]+\.html$}{}s;
    return $base;
}

sub _get_file {
    my $url_or_path = "" . shift; # force to string.

    if ($url_or_path =~ m{^https?://}) {
        return http_get($url_or_path);
    }
    if ($url_or_path =~ m{\.html$}) {
        return path($url_or_path)->slurp_utf8;
    }
    return path($url_or_path)->slurp_raw;
}

sub _get_content {
    my $xhtml = shift;
    if ($xhtml =~ m{/card\d+\.html$}) { # 図書カード
        unless ($xhtml =~ m{^https?://}) { # $xhtml shuld be \d+/card\d+.html
            $xhtml = "$AOZORA_CARDS_URL/$xhtml";
        }
        my $text = _get_file($xhtml);
        my $tree = Aozora2Epub::XHTML::Tree->new($text);
        my $xhtml_url;
        $tree->process('//a[text()="いますぐXHTML版で読む"]' => sub {
            $xhtml_url = shift->attr('href');
        });
        my $xhtml_uri = URI->new($xhtml_url)->abs(URI->new($xhtml));
        return _get_content($xhtml_uri->as_string);
    }
    if ($xhtml =~ m{/files/\d+_\d+\.html$}) { # XHTML
        unless ($xhtml =~ m{^https?://}) { # $xhtml shuld be \d+/files/xxx_xxx.html
            $xhtml = "$AOZORA_CARDS_URL/$xhtml";
        }
        my $text = _get_file($xhtml);
        return ($text, _base_url($xhtml));
    }
    # XHTML string
    return (qq{<div class="main_text">$xhtml</div>}, undef);
}

sub new {
    my ($class, $content, %options) = @_;
    my $self =  bless {
        files => [],
        epub => Aozora2Epub::Epub->new,
        title => undef,
        author => undef,
        bib_info => '',
        notation_notes => '',
    }, $class;
    $self->append($content, %options, title=>'') if $content;
    return $self;
}

sub _cat_url {
    my ($base, $path) = @_;
    unless ($base =~ m{^https?://}) {
        return path($base, $path);
    }
    return URI->new($path)->abs(URI->new($base));
}

sub _build_elemlist_from_xhtml {
    my $xhtml = shift;
    my $tr = Aozora2Epub::XHTML->new_from_string(qq{<div class="main_text">$xhtml</div>});;
    return @{$tr->contents};
}

sub append {
    my ($self, $xhtml_like, %options) = @_;