package Ernad::FullTextFetch;

## not strictly a part of Ernad

use strict;
use warnings;

use Carp qw(confess);
use File::Basename;
use File::Find;

my @bads=('bod','cit','full','hed','raw','txt','cite','abs','body','utf8~',
          'pdf.body','pdf.abs','pdf.cite','pdf.txt.xml','pdf.txt.body','xml',
          'pdf.txt.xml','pdf.txt.utf8','txt.xml','pdf.txt~','txt.utf8',
          'txt-utf8','txt.body');

## a hash of full-text fetch functions
our $fff;

$fff->{'nep'}= sub {
  my $handle=shift;
  my $dir=shift // confess "no directory set";
  my $cut_length=length($dir)+1;
  if(not $handle=~m/repec\:([^\:]+)\:([^\:]+)\:(.*)/i) {
    return "bad handle $handle";
  }
  my $a=$1;
  my $s=$2;
  my $n=$3;
  ## JMBC's magic line
  $n =~ s/[\/\&|\;\(\)]/\_/g;
  my $guessDir = $dir.'/'.$a.'/'.$s.'/'.$n;
  my $guess_dir_short_name=substr($guessDir,$cut_length);
  my @pdfs;
  my $check_stored_file = sub {
    my $file=shift;
    if(-d $file) {
      return 0;
    }
    foreach my $ext (@bads) {
      my $trunc=$file;
      if($trunc=~s/\.\Q$ext\E$//) {
        ## if file without ending exists
        #  if(-f $trunc) {
        #   return 0;
        #  }
        return 0;
      }
    }
    my $bana=basename($file);
    if($bana=~m|^references|) {
      return 0;
    }
    return 1;
  } ;   
  my $wanted_url = sub {
    my $name=$File::Find::name;
    if(&{$check_stored_file}($name)) {
      my $short_name=substr($name,$cut_length);
      push(@pdfs,$short_name);
    }
  } ;
  if( -d $guessDir ) {
    find($wanted_url, $guessDir);
    if(not scalar(@pdfs)) {
      return "no suitable files in $guessDir";
    }
    return \@pdfs;
  }
  else {
    return "no directory $guessDir";
  }
} ;

1;