lib/CXGN/Tools/Tsearch.pm

   1 package CXGN::Tools::Tsearch;
   2 use strict;
   3 use warnings;
   4
   5 =head1 NAME
   6
   7 Functions for working with the Postgres tsearch module used for text indexing
   8
   9 =head1 SYNOPSIS
  10
  11
  12 =head1 DESCRIPTION
  13
  14 =head1 Authors
  15
  16 Naama Menda (nm249@cornell.edu)
  17
  18 =cut
  19
  20
  21 =head2 process_string
  22
  23  Usage: CXGN::Tsearch::process_string('some string' , 1)
  24  Desc:  a class function for preparing a string for matching to a tsvector field
  25  Ret:   a processed string
  26  Args:  string , and optional flag for removing spaces and dots
  27  Side Effects:
  28  Example:
  29
  30 =cut
  31
  32 sub process_string {
  33     my $string = shift;
  34     my $spaces=shift;
  35
  36     if ($spaces) {
  37         print STDERR "process string about to replace spaces and dots for string $string ...\n";
  38         $string =~ s/\s//g;  #replace spaces with nothing
  39         $string =~ s/\.//g;  #replace dots with nothing - important for a multiple synonym string
  40     }
  41
  42     $string =~ s/\(.*?\)//g;
  43     $string =~ s/\s/&/g;
  44     $string =~ s/:/&/g;
  45     $string =~ s/&+/&/g;
  46     $string =~ s/((\`.*?&)|(\'.*?&))/&/g;
  47     $string =~ s/((\`.*?$)|(\'.*?$))//g;
  48     $string =~ s/^&|&$//;
  49     $string =~ s/\(|\)//g;
  50     $string =~ s/\^//g;
  51
  52     #$cvterm_hash->{name} =~ s/(.+?)\s\b(\w*)\/(\w*)\b/$1&$2|$1&$3/;
  53     return $string;
  54 }
  55
  56
  57 =head2 do_insert
  58
  59  Usage: $cvterm->do_insert($match_type, [pub_id, rank, headline] )
  60  Desc:  insert a new cvterm_pub_rank  or a locus_pub_rank
  61  Ret:   number of lines stored in _pub_rank table
  62  Args:  match_type and list of lists
  63  Side Effects: calls the store function
  64  Example:
  65
  66 =cut
  67
  68 sub do_insert {
  69     my $self=shift;
  70     my $match_type=shift;
  71     my $array_ref=shift;
  72     my @array = @$array_ref;
  73     my $count=0;
  74
  75     for my $ref(@array)  {
  76         my $pub_id = @$ref[0];
  77         my $rank= @$ref[1];
  78         my $headline=@$ref[2];
  79
  80         #Check that there is not already a locus-publication link with this match type:
  81         #Look at the match_type of each row where locus_id and pub_id match the current ones.
  82         #If any of these are "name_abstract" then we do not insert our result into the table.
  83         $self->set_pub_id($pub_id);
  84         $self->set_rank($rank);
  85         $self->set_match_type($match_type);
  86
  87         $headline =~ s/\'//g; #gets rid of single quotes from headline text
  88         $self->set_headline($headline);
  89         my $store= $self->store(); #store function should call exists function!
  90         if ($store) { $count++ ; }
  91     }
  92     #if (!$array) { $array=0 ; }
  93     print STDERR "Found " .scalar(@array) . " lines. Inserted $count $match_type rows into _pub_ranking table**\n";
  94     return $count;
  95 }
  96
  97
  98 return 1;