#!perl
use strict;
use Data::Dumper;
use Carp;

#
# This is a SAS Component
#

=head1 text_search


text_search performs a search against a full-text index maintained 
for the CDMI. The parameter "input" is the text string to be searched for.
The parameter "entities" defines the entities to be searched. If the list
is empty, all indexed entities will be searched. The "start" and "count"
parameters limit the results to "count" hits starting at "start".


Example:

    text_search [arguments] < input > output

The standard input should be a tab-separated table (i.e., each line
is a tab-separated set of fields).  Normally, the last field in each
line would contain the identifer. If another column contains the identifier
use

    -c N

where N is the column (from 1) that contains the subsystem.

This is a pipe command. The input is taken from the standard input, and the
output is to the standard output.

=head2 Documentation for underlying call

This script is a wrapper for the CDMI-API call text_search. It is documented as follows:

  $return = $obj->text_search($fids)

=over 4

=item Parameter and return types

=begin html

<pre>
$fids is a fids
$return is a reference to a hash where the key is a fid and the value is an annotations
fids is a reference to a list where each element is a fid
fid is a string
annotations is a reference to a list where each element is an annotation
annotation is a reference to a list containing 3 items:
	0: a comment
	1: an annotator
	2: an annotation_time
comment is a string
annotator is a string
annotation_time is an int

</pre>

=end html

=begin text

$fids is a fids
$return is a reference to a hash where the key is a fid and the value is an annotations
fids is a reference to a list where each element is a fid
fid is a string
annotations is a reference to a list where each element is an annotation
annotation is a reference to a list containing 3 items:
	0: a comment
	1: an annotator
	2: an annotation_time
comment is a string
annotator is a string
annotation_time is an int


=end text

=back

=head2 Command-Line Options

=over 4

=item -c Column

This is used only if the column containing the subsystem is not the last column.

=item -i InputFile    [ use InputFile, rather than stdin ]

=back

=head2 Output Format

The standard output is a tab-delimited file. It consists of the input
file with extra columns added.

Input lines that cannot be extended are written to stderr.

=cut


my $usage = "usage: text_search [-start N] [-count N] [-entity name -entity name ..] search-string\n";

use Bio::KBase::CDMI::CDMIClient;
use Bio::KBase::Utilities::ScriptThing;

my $start = 0;
my $count = 100;
my @entities;
my $kbO = Bio::KBase::CDMI::CDMIClient->new_for_script('start=s' => \$start,
				     'count=s' => \$count,
				     'entity=s' => \@entities);

if (@ARGV == 0 || ! $kbO) { print STDERR $usage; exit }

my $string = join(" ", @ARGV);

my $res = $kbO->text_search($string, $start, $count, \@entities);
for my $entity (keys %$res)
{
    my @hdr;
    for my $hit (@{$res->{$entity}})
    {
	my($weight, $data) = @$hit;
	if (!@hdr)
	{
	    @hdr = sort keys %$data;
	    print join("\t", "#", @hdr), "\n";
	}
	print join("\t", $entity, @$data{@hdr}), "\n";

    }
}