crayz.pl

sub parse {

    my $self = shift;
    my $text = lc shift;
    my $ret  = shift;

    $ret->{fixed} ||= 0;
    return 'overflow_error' if $ret->{fixed} > 4;    # nunca deve ocorrer, mas bugs sempre estao soltos.

    sub reformate_espaces {
        my $text = shift;
        $text =~ s/\s+/ /go;                         # espacos duplicadoss
        $text =~ s/^\s+//o;                          # trim
        $text =~ s/\s+$//o;

        $text =~ s/\s*\(\s*/ ( /go;                  # trim entre os parenteses
        $text =~ s/\s*\)\s*/ ) /go;

        $text =~ s/(\w)\)/$1 ) /go;                  # untrim entre as palavras e os parenteses
        $text =~ s/\((\w)/( $1/go;

        return $text;
    }

    #fala "text = $text";
    my $tag      = qr/!?[a-z]+/;
    my $operador = qr/[&\|]/o;

    $text =~ s/\(\s+\)//go;
    $text =~ s/\(($operador\s*)+\)//go;
    $text = reformate_espaces($text);

    # operadores repetidos
    $text =~ s/($operador)(:?\s?$operador)+/$1/g;

    # trocando palavras sem operadores por "e"
    my $reg = qr /($tag+)\s+([^\&\|])?($tag+)/o;
    while ( $text =~ /$reg/ ) {
        $text =~ s/$reg/$1 & $2$3 /o;
        $text = reformate_espaces($text);
    }

    # coloca "e" entre parentes depois
    $reg = qr /\s\)\s($tag)/o;
    $text =~ s/$reg/ ) & $1/o while $text =~ /$reg/;

    # ou antes
    $reg = qr /($tag)\s\(/o;
    $text =~ s/$reg/$1 & (/o while $text =~ /$reg/;

    # reformata, afinal, ninguem sabe que bagunca pode estar o texto nessas alturas!
    $text = reformate_espaces($text);

    $text =~ s/\s*($operador)\s*/ $1 /go;

    # --------------------------------------
    # tratando erros de sintax.

    # operadores repetidos no final
    $text =~ s/$operador+\s*$//o;
    $text =~ s/^\s*$operador+//o;

    $text =~ s/\(\s*$operador+/(/go;
    $text =~ s/$operador+\s*\)/)/go;

    my $open  = () = $text =~ /\(/g;
    my $close = () = $text =~ /\)/g;
    if ( $open != $close || $text =~ /\(\s+\)/ ) {

        $ret->{fixed}++;
        $text =~ s/[\&\|\(\)!]/ | /g;

        $text = $self->parse( $text, $ret );
    }

    $text = reformate_espaces($text);

    return $text;
}

sub get_query {
    my ( $self, $text ) = @_;
    my $ret = {};

    $text = $self->parse( $text, $ret );

    return { tratado => $text, %$ret };

}

sub parse_search {
    my $self = shift;
    my $opt  = shift;

    my $keywords       = {};
    my $reverse_search = $self->reverse_search();
    my $translate      = $self->translate();

    my $text = lc $converter->convert( $opt->{text} );


    $text =~ s/\b(ou|or)\b/ | /go;       # traduzindo "ou" por |
    $text =~ s/\b(e|and|\+)\b/ & /go;    # traduzindo "e" por &

    $text =~ tr/\[\]/()/;
    $text =~ s/\b(não|nao|no)(\s+|$)/ !/go;    # adicioando NOT
    $text =~ s/\bpeitos?\b/ !pussy /gi;

    $text =~ s/[^a-z\&\+\|\(\)\[\]\s\!0-9]//go;
    # operador nao nao = sim =D
    $text =~ s/!!//g;

    foreach my $eng ( sort { length $b <=> length $a } keys %$reverse_search ) {
        my $find = lc $eng;
        $find =~ s/s$//;
        $text =~ s/${find}s?/ $reverse_search->{$eng} /gi;
    }
    $text =~ s/!\s+/!/go;

    my @words = split /\s+/, $text;
    my @out = ();
    foreach my $w (@words) {
        if ( $w =~ /^!?[\|&()]$/ || ( $w =~ /^!(.+)/ && exists( $translate->{$1} ) ) ) {
            push @out, $w;
            next;
        }
        if ( exists( $translate->{$w} ) ) {
            push @out, $w;
        }
    }
    $text = join ' ', @out;

    my $ret = $self->get_query($text);

    $ret->{tsquery} = $ret->{tratado};

    $ret->{text} = $self->_translate_tsquery( $ret->{tsquery} );

    return $ret;
}

sub _translate_tsquery {
    my $self    = shift;
    my $tsquery = shift;

    my $translate = $self->translate();

    $tsquery =~ s/ \| / ou /go;
    $tsquery =~ s/ & / e /go;
    $tsquery =~ s/!/não /go;

    foreach ( keys %$translate ) {
        $tsquery =~ s/\b$_\b/$translate->{$_}/g;
    }
    $tsquery =~ s/\s+/ /go;

    return $tsquery;
}