#!/usr/bin/perl

use strict;
use warnings; 
use DBI;
use PDF::API2;
use Data::Dump 'pp';
use File::Basename;
use File::Find::Rule;

my $t0 = time();

#my $pdfs = get_pdfs();

my $pdfs = ();
%{$pdfs} = (
 "testa.pdf" => 0,
 "testb.pdf" => 0,
);

#pp $pdfs;

my @search_dirs = ('/doca', '/docb');

my $rule =  File::Find::Rule->new;
$rule->file;
$rule->name( '*.pdf' );

my $count = 0;
my $totalpages = 0;

for my $file ($rule->in(@search_dirs)){
  
  my ($filename, $path) = fileparse($file);
  #print " *$filename* ^$pdfs->{$filename}^\n";

  if (exists $pdfs->{$filename}){
      
    my $pages = get_pagecount($file);
    $totalpages += $pages;
    print " $filename | $pages | $path\n"; # print to file
    ++$count;
    
  }#else { print " File $filename not found in $path\n";}
  
};

my $dur = time()-$t0;
print "\n\n Results:  $count files found in $dur seconds - Total number of pages: $totalpages\n\n\n";

sub get_pagecount {
    
  my $doc = shift;
  my $pages;
  
  eval { my $pdf = PDF::API2->open($doc);
    $pages = $pdf->pages;
  };
  
  if ($@){
    warn "$doc | Error captured : $@\n";
  } else {
    return $pages;
  };
  
}

sub get_pdfs {
    
  my $dbh = DBI->connect ("dbi:CSV:", undef, undef, {
      f_ext   => ".txt/r",
      #f_dir  => "db",
      f_enc   => "utf-8",
      RaiseError => 1 }
  ) or die "Cannot connect: $DBI::errstr";
         
  $dbh->{csv_tables}{prod_pdf_files} = {
    f_file    => "prod_pdf_test.txt", # list of file names to search - test file = prod_pdf_test.txt
    col_names => [qw( doc_name file_name acc_nbr )],
  };

  my $sth = $dbh->prepare ("
      SELECT DISTINCT file_name 
      FROM prod_pdf_files 
      WHERE LOWER(file_name) LIKE ?
   ");
  $sth->execute('%.pdf');
  my $pdfs = $sth->fetchall_arrayref();
  my %pdfs = map{lc($_->[0]) => 0} @$pdfs;
  #pp  \%pdfs;
  return \%pdfs;
}