package PubMed::Record;

use strict;
use warnings;

use Carp;

my $date_paths;
$date_paths->[0]='/MedlineCitation/Article/Journal/JournalIssue/PubDate/Year';
$date_paths->[1]='/MedlineCitation/Article/Journal/JournalIssue/PubDate/MedlineDate';
my $pmid_path='/MedlineCitation/PMID';

## a counter to see how sucessful the paths are
our $count_paths;
foreach my $date_path (@$date_paths) {
  $count_paths->{$date_path}=0;
}

sub get_year {
  my $txt=shift;
  my $doc=shift;
  my $xpc = XML::LibXML::XPathContext->new($doc);
  my $year=0;
  foreach my $path (@{$date_paths}) {
    $year=try_to_find_year($xpc,$path,$txt) or next;
    if($year) {
      $count_paths->{$path}++;
      last;
    }
  }
  if(not $year and $main::test) {
    print "no year in\n$txt\n";
    exit;
  }
  if(not $year=~m|^\d{4}$|) {
    confess "I have a bad year $year in\n$txt\n";
  }
  #print "year is $year.\n";
  return $year;
  #print $txt;
}

sub get_pmid {
  my $txt=shift;
  my $doc=shift;
  my $xpc = XML::LibXML::XPathContext->new($doc);
  my @nodes=$xpc->find($pmid_path)->get_nodelist;
  if(not scalar @nodes) {
    confess "I found no pmid in \n$txt.";
  }
  if(@nodes) {
    if(scalar(@nodes) > 1) {
      confess "I found several years in \n$txt;";
    }
  }
  my $pmid=$nodes[0]->textContent;
  if(not $pmid=~m|^\d+$|) {
    confess "This '$pmid' is not a pmid";
  }
  return $pmid;
}


sub try_to_find_year {
  my $xpc=shift;
  my $path=shift;
  my $txt=shift;
  my @nodes=$xpc->find($path)->get_nodelist;
  if(@nodes) {
    if(scalar(@nodes) > 1) {
      confess "I found several years in \n$txt;";
    }
    my $year=substr($nodes[0]->textContent,0,4);
    return $year;
  }
  return '';
}

1;