#!/usr/bin/perl

#####!C:\Perl64\bin\perl.exe
use Time::Local; 

# CabanissCheck
# goes through scans directories for EAD linking
# picks up box and folder numbers from directory names
# checks to see that all tiffs in those directories match those box and folder numbers
# checks for missing sequence numbers

#  expected structure:
#  Collection_dir
#     Scans.*?_Box_\d{1,4}
#        Folder_\d{1,4}
#           u0003_0000252_BBFF\d{3}(.tif)?  or is subdir
#               u0003_0000252_BBFF\d{3}_\d{3}.tif

#  where BB is box number, FF is folder number and tiffs are in another subdirectory only if there are pages
#  check sequence on item and page level

# 

# jody DeRidder, 11/12/09

&timestamp;

# print $timestamp."\n";

$output = "S:\\Digital Projects\\Administrative\\scripts\\output\\CabanissTest_".$timestamp.".txt";
$thiscoll= "u0003_0000252";

open (OUT, ">".$output) or die "can't open $output\n";

print "Do you want to check Cabaniss content in \n1) the Complete folder or \n2) the In Progress folder?\n";
print "Please choose a number, then hit enter\n";
$answ = <STDIN>;
if ($answ =~ /1/){ $base = "S:\\Digital\ Projects\\Digital_Coll_Complete\\";}
elsif ($answ =~ /2/){$base = "S:\\Digital\ Projects\\Digital_Coll_in_progress\\";}
else { print "Sorry, I didn't get that.  Exiting now.  Please start over. \n";  sleep(2);  exit;}

opendir (BASE, $base) or die "can't look through $base\n";
while ($file = readdir(BASE)){
    if ($file =~ /^\./){next;}  # skip dot files
    $path = $base.$file;
    if ($file =~ /u0003_0000252/ && -d $path){
       $target = $path;
        }
     }
if (! $target){ print "Unable to locate Cabaniss folder in $base.  \n  Sorry!!\n\n";  sleep(2);  exit;}   
      
print "Results will be in the $output file....\n  when you see a \"Good bye!\" and this screen closes.\n";
print "Don't open it yet!  :-)  \n\n";
print ". . . working . . .  (be patient) . . . \n\n";

push (@mydirs, $target."\\");

foreach $dir (@mydirs){


  opendir(DIR, $dir) or die "can't read through $dir directory\n";
  while ($file = readdir(DIR)){
    if ($file =~ /^\./ || $file =~ /Admin/){
      next;  # skip dot files
      }
    $path = $dir.$file;
    if (-d $path){
       if ($file =~ /^([a-z]{1}\d{4}\_\d{7})\_(\d{7})$/){
          $collnum = $1;
          $itemID = $1."_".$2;
          $thisItem = $2;
          if ($thisItem =~ /^(\d{2})(\d{2})(\d{3})/){
             $boxnum = $1;  
             $foldernum = $2;
             $item_sequence = $3 + 0; # gets rid of padding zeros
             $bf = $boxnum."_".$foldernum;
             push (@{$ItemSeq{$bf}}, $item_sequence);

           #  print OUT "$boxnum $foldernum $item_sequence  $file\n";
             }
          if ($collnum ne $thiscoll){ push (@badform, "$file does not match collection number $thiscoll\n");}  
          }
        push (@mydirs, $path."\\");
        }

# not checking subpages
    elsif ($file =~ /^([a-z]{1}\d{4}\_\d{7})\_(\d{7})(\_(\d{4}))?\.tif$/){
       $collnum = $1;
       $itemID = $1."_".$2;
       $thisItem = $2;
       $thisPage = $4 +0 ;
       if ($4){ $apage = 1;}
       else{ undef $apage;}

       if ($collnum ne $thiscoll){ push (@badform, "$file does not match collection number $thiscoll\n");}

     #  print "collnum $collnum, item $itemID this Item $thisItem, this page $thisPage\n";
       if ($thisItem =~ /^(\d{2})(\d{2})(\d{3})/){
          $boxnum = $1;  
          $foldernum = $2;
          $item_sequence = $3 + 0; # gets rid of padding zeros
          }
       else{  push (@badform, "Can't get boxnum, foldernum, item sequence from $thisItem");}

      $bf = $boxnum."_".$foldernum;
      if (! $apage){
         # print OUT "$boxnum $foldernum $item_sequence  $file\n";
          push (@{$ItemSeq{$bf}}, $item_sequence);
          } # this catches tifs not in subdirectories



       $tifcount ++;
       push (@{$collnumItems{$collnum}{$file}}, $dir);   # keeps list of directories where items are found
       undef $box;
       undef $folder;
       
       ($checkdir = $dir) =~ s,.*?(u0003\_0000252),u0003\_0000252,;  # pull off first part of this directory name

      # what is the series number and folder number?
        if ($checkdir =~ /Box\_(\d{1,2})(_[^\\]*)?\\Folder\_(\d{1,2})(_[^\\]*)?\\([^\\]+\\)?/i){
              $box = sprintf ("%02d", $1);  # left pad if needed
              $folder = sprintf ("%02d",$3);
              $itemdir = $5;
              if (($box ne $boxnum) || ($folder ne $foldernum)){ push(@badform, "$file  name does not reflect Box $box Folder $folder at $path");}
              $itemdir =~ s,\\,,g;
              
              if ($itemdir){
                if (!$thisPage){push (@errors, "$file is in $itemdir directory, but has no page number\n");}
                elsif ($itemdir ne $itemID){ push (@wrongdir, "$file is in $itemdir;  no match on item id\n");}
                else{  push (@{$pages{$itemdir}}, $thisPage );}   # get page number sequence without the padding
                }
              }
          else{  push (@errors, "cannot parse out box and folder from $path\n");}
          }
       else{ 
         if ($file =~ /thumbs\.db/i  || $path =~ /Admin/){ next;}
         push (@badform, "$file in $checkdir");
         } 
       }
   close(DIR);
   }
$subtract = 0;
@allfiles = sort by_number (keys (%{$collnumItems{$collnum}}));
foreach $file (@allfiles){
#   print "found $file\n";
   @mydirs = @{$collnumItems{$collnum}{$file}};
   $numdirs = scalar (@mydirs);
   if ($numdirs > 1){
     $collect = join ("   \n", @mydirs);
     push (@dupes, "$file exists in more than one place:\n $collect");
     $subtract += $numdirs -1;  # subtract from total count of files
     }
   }


@boxfolders = sort keys(%ItemSeq);
foreach $bf (@boxfolders){
   ($box, $folder) = split ("_", $bf);
  # print "Box $box   Folder $folder";
   @myItems = sort by_number( @{$ItemSeq{$bf}});
   $numItems = scalar (@myItems);
   push (@foldercount, "Box $box   Folder $folder contains $numItems items");
   $totalItems += $numItems;
   for ($i = 1;  $i <= $numItems; $i++){
     if ($myItems[$i-1] ne $i){
       push (@badcount, "In Box $box Folder $folder: expected sequence $i instead of ".$myItems[$i-1]);
       }
     }
   }
@hasPages = sort keys(%pages);
foreach $doc (@hasPages){
   @mypages = sort by_number( @{$pages{$doc}});
   $numItems = scalar (@mypages);
   push (@pagecounts, "$doc has  $numItems pages");
   for ($i = 1;  $i <= $numItems; $i++){
     if ($mypages[$i-1] ne $i){
       push (@badcount, "$doc page problem:  expected sequence $i instead of ".$mypages[$i-1]);
       }
     }
   }
$all = (scalar @allfiles) - $subtract;
print OUT "\nTotal count of valid content in this upload:\n  $totalItems items and $all digital files\n";
#print "tiffcount $tifcount\n";





if (@badform || @badcount || @wrongdir || @errors){
     print OUT "\nTROUBLE: \n";
     print OUT "-------------------------------------------\n";
     }
 else{
     print OUT "All is GREAT!!  GOOD WORK!!  :-) \n";
    }
  
   if (@errors){  
        print OUT "\nThese general errors impeded this script, so everything may NOT have been checked:\n";
        print OUT "Please repair and run this again.\n";
        foreach (@errors){ print OUT "    $_\n";}
        } 
   if (@wrongdir){
        print OUT "\nThe following files or directories do NOT reflect the name of their parent directory\n";
        foreach (@wrongdir){ print OUT "    $_\n";}
        }

   if (@badform){
        print OUT "\nThe following filenames or directories are not in the correct format:\n";
        foreach (@badform){ print OUT "    $_\n";}
        }

    if (@badcount){
        print OUT "\nThe following folders or items have sequence problems:\n";
        foreach (@badcount){ print OUT "    $_\n";}
        }


print OUT "\n\nHere's some more info:\nFOLDER COUNTS:\n_________________\n";

foreach (@foldercount){ print  OUT $_."\n";}
print OUT "\n\nPAGE COUNTS:\n_______________________\n";
foreach (@pagecounts){ print OUT $_."\n";}
 
close OUT;


print "Good bye!\n";
exit;



sub by_number {$a <=> $b;}

sub timestamp{
print "hit enter twice please\n";
$date = `date`;
$time = `time`;

#print $date."\n";
if ($date =~ /.*? (\d*)\/(\d*)\/(\d*)/){
     $date = $3.$1.$2;
      }
#print $date."\n";
if ($time =~ /.*? (\d*)\:(\d*)\:(\d*)\./){
     $time = $1.$2.$3;
     }
#print $time."\n";
$timestamp = $date."T".$time;
#print $timestamp."\n";

# following for unix
#($sec, $min, $hour, $mday, $mon, $year, $wday, $yday, $isdst) = gmtime($mydate);
#$mon ++;
#if ($mon < 10){ $mon="0".$mon;}  #need 2 digits
#if ($sec < 10){ $sec="0".$sec;}
#if ($min < 10){ $min="0".$min;}
#if ($hour < 10){ $hour="0".$hour;}
#if ($mday < 10){ $mday="0".$mday;}

#$year = $year + 1900;
#$timestamp= "$year-$mon-$mday\T$hour:$min:$sec\Z";
}