package PubMed::Api;

use strict;
use warnings;

use Carp qw(confess);
use Encode;
use Math::Round;
use File::Slurp;
use File::Touch;
use LWP::Simple;
use XML::LibXML;

use PubMed::Common;
use PubMed::Files;
use PubMed::Paths;

## assemble the esearch URL
my $base='http://eutils.ncbi.nlm.nih.gov/entrez/eutils';
my $general='db=pubmed&retmode=xml';
my $get_id="efetch.fcgi?$general&id";
my $dt='datetype=edat&retmax=99999';
my $sb="esearch.fcgi?$general&term=publisher+%5Bsb%5D&$dt";

## a search resulting in string. Wait is optional
sub get_from_api {
  my $ids=shift // confess "I need ids arrref here.";
  my $wait=shift // '';
  my $in_ids=join(',',@{$ids});
  my $url="$base/$get_id=$in_ids";
  #print "$url\n";
  my $output=get($url) or return;
  my $bytes=encode_utf8($output);
  if($wait) {
    my $count=scalar @{$ids}+1;
    my $seconds=nearest(1,log($count));
    sleep $seconds;
  }
  return $bytes;
}

## a search resulting in a file. Wait is done by caller
sub sb_search {
  my $date=shift // confess "I need a date here.";
  if(not $date=~m|\d{4}/\d{2}/\d{2}|) {
    confess "This is a bad date: $date";
  }
  my $file=shift // '';
  my $url="$base/$sb&mindate=$date&maxdate=$date";
  my $output=get($url) or return;
  my $bytes=encode_utf8($output);  
  if($file) {
    write_file($file, $bytes);
  }
  return $bytes;
}

sub pmapi_file_to_fapi {
  my $pmapi_file=shift // confess "I need a file argument here.";
  if(not -f $pmapi_file) {
    confess "I can see the file $pmapi_file";
  }
  my $doc=&PubMed::Common::load_and_return_xml($pmapi_file) or die;
  foreach my $pmart ($doc->getElementsByTagName('PubmedArticle')) {
    &save_pmart($pmart);
  }
}

sub pmapi_bytes_to_fapi {
  my $bytes=shift;
  my $date=shift // '';
  my $touch_date=shift // '';
  my $doc;
  eval {
    $doc = XML::LibXML->load_xml(string => (\$bytes));
  };
  if(ref $doc ne 'XML::LibXML::Document') {
    confess "I could not parse file '$bytes'\nLibXML says $@\n";
  }
  foreach my $pmart ($doc->getElementsByTagName('PubmedArticle')) {
    &save_pmart($pmart,$date,$touch_date);
  }
}

sub save_pmart {
  my $pmart=shift;
  my $date=shift // '';
  my $touch_date=shift // '';
  my $ref_pmart = ref($pmart);
  if(not $ref_pmart eq 'XML::LibXML::Element') {
    confess "You did not give me an element, but a $ref_pmart.";
  }
  my $pmid=&PubMed::Common::get_pmid($pmart);  
  my $file=&PubMed::Files::pmid_to_fapi($pmid);
  write_file($file,{binmode => ':utf8'},$pmart->toString(1));
  if($date and $touch_date) {
    ## append to infa file
    my $out_file=$PubMed::Paths::infa_dir.'/'.$date;
    open(F,">> $out_file"); 
    print F "$pmid\n";
    close F;
    $touch_date->touch($out_file);
  }
}

## old function that tries to use fapi
#sub get_purd_by_pmid {
#  my $pmid=shift;
#  ## set a date on the file
#  my $date=shift; 
#  my $do_i_check=shift // '';
#  my $fapi=&PubMed::Files::pmid_to_fapi($pmid);
#  if(not -f $fapi) {
#    return '';
#  }
#  my @lines=read_file($fapi);
#  my $in_record=0;
#  my $purd='';
#  foreach my $line (@lines) {
#    if($line=~m|^\s+<MedlineCitation|) {
#      $purd.=$line;
#      $in_record=1;
#      next;
#    }
#    if($in_record) {
#      if($line=~m|^\s+</MedlineCitation|) {
#        $purd.=$line;
#        $in_record=0;
#        next;
#      }
#      $purd.=$line;
#    }
#  }
#  return $purd;
#}