File Coverage

blib/lib/LyricFinder/Musixmatch.pm

Criterion	Covered	Total	%
statement	23	97	23.7
branch	0	30	0.0
condition	0	30	0.0
subroutine	8	11	72.7
pod	2	2	100.0
total	33	170	19.4

package LyricFinder::Musixmatch;

use strict;
use warnings;
use Carp;
use HTML::Strip;
use URI::Escape;
use parent 'LyricFinder::_Class';

our $haveLyricsCache;
BEGIN {
	$haveLyricsCache = 0;
	eval "use LyricFinder::Cache; \$haveLyricsCache = 1; 1";
}

my $Source = 'Musixmatch';
my $Site   = 'https://www.musixmatch.com';
my $DEBUG  = 0;

sub new
{
	my $class = shift;

	my $self = $class->SUPER::new($Source, @_);
	@{$self->{'_fetchers'}} = ($Source);
	unshift(@{$self->{'_fetchers'}}, 'Cache')  if ($haveLyricsCache
			&& $self->{'-cache'} && $self->{'-cache'} !~ /^\>/);

	bless $self, $class;   #BLESS IT!

	return $self;
}

sub fetch {
	my ($self, $artist_in, $song_in) = @_;

	$self->_debug("Musixmatch::fetch($artist_in, $song_in)!");

	return ''  unless ($self->_check_inputs($artist_in, $song_in));
	return ''  if ($self->{'Error'} ne 'Ok');

	# first, see if we've got it cached:
	$self->_debug("i:haveCache=$haveLyricsCache= -cachedir=".$self->{'-cache'}."=");
	if ($haveLyricsCache && $self->{'-cache'} && $self->{'-cache'} !~ /^\>/) {
		my $cache = new LyricFinder::Cache(%{$self});
		if ($cache) {
			my $lyrics = $cache->fetch($artist_in, $song_in);
			if (defined($lyrics) && $lyrics =~ /\w/) {
				$self->_debug("..Got lyrics from cache.");
				$self->{'Source'} = 'Cache';
				$self->{'Site'} = $cache->site();
				$self->{'Url'} = $cache->url();

				return $lyrics;
			}
		}
	}

	$self->{'Site'} = $Site;

	(my $artist = $artist_in) =~ s#\s*\/.*$##;  #ONLY USE 1ST ARTIST, IF MORE THAN ONE!
	(my $song = $song_in) =~ s#\s*\/\s*#\-#;  #FIX SONGS WITH "/" IN THEM!
	$artist =~ s/\&/feat/;    #DON'T ASK ME WHY "&" => "feat"? BUT IT DOES!
	$song =~ s/\&/feat/;
	$artist =~ s/ +/\-/g;
	$song =~ s/ +/\-/g;
	$artist =~ tr/A-Z/a-z/;
	$song =~ tr/A-Z/a-z/;
	$artist = uri_escape_utf8($artist);
	$song = uri_escape_utf8($song);
	$artist =~ s/[^a-z0-9\-\%]//gi;
	$song   =~ s/[^a-z0-9\-\%]//gi;

	# Their URLs look like e.g.:
	#https://www.musixmatch.com/lyrics/Artist-name/Title
	$self->{'Url'} = $Site . '/lyrics/'
			. join('/', $artist, $song);

	my $lyrics = $self->_web_fetch($artist_in, $song_in);
	if ($lyrics && $haveLyricsCache && $self->{'-cache'} && $self->{'-cache'} !~ /^\
		$self->_debug("=== WILL CACHE LYRICS! ===");
		# cache the fetched lyrics, if we can:
		my $cache = new LyricFinder::Cache(%{$self});
		$cache->save($artist_in, $song_in, $lyrics)  if ($cache);
	}
	return $lyrics;
}

sub _parse {
	my $self = shift;
	my $html = shift;

	$self->_debug("Musixmatch::_parse()!");
	my $goodbit = '';
	my $text = '';
	my $mm_status = '';
	my $hs   = HTML::Strip->new();

	#METHOD 1:  MERGE LYRICS FROM BOTH (OR MORE?) LYRIC "SPANS" MM LIKES TO BREAK LYRICS UP INTO
	#2 SEPARATE SPAN TAGS:  A SHORT (ABBREVIATED) PART, FOLLOWED BY THE REST OF THE LYRICS:
	while ($html =~ s#\(.+?)\<\/span\>##s) {
		$mm_status = $1;
		$goodbit = $2;
		$text .= $hs->parse($goodbit) . "\r\n";
	}
	unless ($text =~ /[a-z]/i) {  #IF METHOD 1 FAILS, THE FULL LYRICS ARE OFTEN FOUND IN THE FIRST "body":
		$html =~ s#\\\"#\x02#gs;  #PROTECT ESCAPED QUOTES
		if ($html =~ m#\"body\"\:\"([^\"]+)#s) {
			$mm_status = 'error!';
			($goodbit = $1) =~ s#\x02#\"#gs;
			$text .= $hs->parse($goodbit);
			$text =~ s#\\n#\n#gs;  #line-feeds are encoded literally in the JSON block.
		}
	}
	if ($text =~ /[a-z]/i) {
		$text .= "\r\n(Musixmatch status: $mm_status)"
				unless ((defined($self->{'-noextra'}) && $self->{'-noextra'}) || $mm_status !~ /\S/);

		#WHILE WE'RE AT IT, SEE IF WE HAVE A COVER IMAGE?!:
		if ($html =~ m#\(.+?)\<\/div\>#s) {
			my $imgdiv = $1;
			if ($imgdiv =~ m#\
				my $imgurl = $1;
				$imgurl = 'https:' . $imgurl  if ($imgurl =~ m#^//#);
				$self->{'image_url'} = $imgurl;
			}
		}

		return $self->_normalize_lyric_text($self->_html2text($text));
	} else {
		carp($self->{'Error'} = "e:$Source - Failed to identify lyrics on result page.");
		return '';
	}
}

1

__END__