view release on metacpan or  search on metacpan

lib/Alvis/QueryFilter.pm  view on Meta::CPAN

use CQL::Parser;
use URI::Escape;

use strict;

our $VERSION = '0.3';
our $verbose = 0;

my ($ERR_OK,
    $ERR_CQL_PARSER_INST,
    $ERR_XML_PARSER,
    $ERR_NO_QUERY,
    $ERR_CQL_PARSE,
    $ERR_NONIMP_NODE_TYPE,
    $ERR_NO_SEQ_LIST,
    $ERR_TREETAGGER,
    $ERR_LEMMA_DICT,
    $ERR_APPLYING_TERM_NE,
    $ERR_APPLYING_TYPING,
    $ERR_APPLYING_ONTO,
    $ERR_STRUCT2XML,
    $ERR_CREATING_CAT_LIST,
    $ERR_CREATING_SEQ_LIST,
    $ERR_CREATING_CQL_TAIL,
    $ERR_NEED_BOTH_TERM_AND_NE
    )=(0..16);

my %ErrMsgs=($ERR_OK=>"",
	     $ERR_CQL_PARSER_INST=>"Instantiating CQL::Parser failed.",
	     $ERR_XML_PARSER=>"Instantiating the XML parser failed.",
	     $ERR_NO_QUERY=>"No query.",
	     $ERR_CQL_PARSE=>"CQL parsing failed.",
	     $ERR_NONIMP_NODE_TYPE=>'Non-implemented CQL node type.',
	     $ERR_NO_SEQ_LIST=>
			      "No current data structure representing the " .
			      "query expansion. No preceding UI2Zebra() " .
			      "call or we're out of sync. You'll get your " .
			      "money back.",
	     $ERR_TREETAGGER=>"Applying the treetagger failed.",
	     $ERR_LEMMA_DICT=>"Applying the lemma dictionary failed.",
	     $ERR_APPLYING_TERM_NE=>"Applying the terms and NEs failed.",
	     $ERR_APPLYING_TYPING=>"Applying the typing rules failed.",
	     $ERR_APPLYING_ONTO=>"Applying the ontology mappings failed.",
	     $ERR_STRUCT2XML=>"Converting the data structure into XML failed.",
	     $ERR_CREATING_CAT_LIST=>
	     "Extracting category part of query failed.",
	     $ERR_CREATING_SEQ_LIST=>
	     "Creating the list of possible expansions failed.",
	     $ERR_CREATING_CQL_TAIL=>"Converting the data structure to a CQL " .
	     "tail failed.",
	     $ERR_NEED_BOTH_TERM_AND_NE=>"We need both a term and a NE dictionary."
	     );

sub new
{
    my $proto=shift;
 
    my $class=ref($proto)||$proto;
    my $parent=ref($proto)&&$proto;
    my $self={};
    bless($self,$class);

    $self->_set_err_state($ERR_OK);

    $self->_init(@_);

    $self->{CQLParser}=new CQL::Parser();
    if (!defined($self->{CQLParser}))
    {
	$self->_set_err_state($ERR_CQL_PARSER_INST);
	return undef;
    }

    $Alvis::Treetagger::verbose = $verbose;
    &Alvis::Treetagger::reopen();
    $self->{termMaxLen} = 0;
    $self->{textFields} = "text";
    $self->{tcanon} = \&canonise_def;
    $self->{ncanon} = \&canonise_def;
    #  this must match the lemma indexing rules in alvis2index.xsl
    $self->{lemmaSearch} = "^[VNJ]";


    return $self;
}

sub _init
{
    my $self=shift;

    $self->{keepLemmas}=1;

    if (defined(@_))
    {
        my %args=@_;
        @$self{ keys %args }=values(%args);
    }

}

sub _set_err_state
{
    my $self=shift;
    my $errcode=shift;
    my $errmsg=shift;

    if (!defined($errcode))
    {
	confess("set_err_state() called with an undefined argument.");
    }

    if (exists($ErrMsgs{$errcode}))
    {
	if ($errcode==$ERR_OK)
	{
	    $self->{errstr}="";
	}
	else
	{
	    $self->{errstr}.=" " . $ErrMsgs{$errcode};
	    if (defined($errmsg))