#!/usr/bin/perl

use Time::Local; 

# testIncoming
# locate all the bad filenames
# things in the wrong directory
# things missing
# bad sequences
# jody DeRidder, 8/25/09

&timestamp;

print $timestamp."\n";

$output = "./output/IncomingTest_".$timestamp.".rtf";

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


$base = "/srv/deposits/content/";

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 (-d $path){
    push (@dirs, $file."/");
    }
  }
close(BASE);


foreach $topdir (@dirs){

  undef (@mydirs);
  $mybase = $base.$topdir."/";
  $thisdir = $mybase;


  opendir(THIS, $thisdir) or die "can't open $thisdir\n";
  while ($file = readdir(THIS)){
    if ($file =~ /^\./){next;}  # skip dot files
       $path = $thisdir.$file;
       if (-d $path){
       push (@mydirs, $path."/");
       }
    }
  close(THIS);


# test here to see if this directory has filename in it...
# if not, ask for file name structure to test

  if ($thisdir =~ /([a-z]{1}\d{4}\_\d{7}(\_\d{7})?)/){
     $coll = $1;
     print "Looking at $thisdir: Is this the collection identifier?   -->$coll\n";
     print "Y or N, and press enter\n";
     $answer = <STDIN>;
     while ($answer =~ /n/i){ 
     undef $coll;
     print "Please enter the collection identifier\n";
     $coll = <STDIN>;
     chop $base;
     print "Is this the collection identifier?   -->$coll\n";
     print "Y or N, and press enter\n";
     $answer = <STDIN>;
     }
    
  } 
  if (!$coll){
  print "Please enter the collection identifier\n";
     $coll = <STDIN>;
     chop $base;
     print "Is this the collection identifier?   -->$coll\n";
     print "Y or N, and press enter\n";
     $answer = <STDIN>;
     while ($answer =~ /n/i){
       undef $coll;
       print "Please enter the collection identifier\n";
       $coll = <STDIN>;
       chop $base;
       print "Is this the collection identifier?   -->$coll\n";
       print "Y or N, and press enter\n";
       $answer = <STDIN>;
       }
    }
 
  print "\nWe will test the files against collection identifier $coll.\n\n";
  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";

  foreach $dir (@mydirs){

#HERE:  do a section for Admin, and one for Metadata

    $count = 1;    # checking sequence count of files in directories

    opendir (DIR, $dir) or die "can't read files in $dir\n";
    while ($adir = readdir(DIR)){
       if ($adir =~ /^\./ || $adir =~ /Thumbs\.db/i){next;}  #skip dot files and thumbs.db
        
       # print "looking at $adir\n";
       ($short = $dir) =~ s,\/srv\/deposit\/content\/,,;
       ($parent = $short) =~ s,.*\/,,;   # parent directory is what follows the last slash, yes?
                                                              # we want to match for this
       $thispath = $dir."/".$adir;
      
       if ($adir =~ /Scans/i || $adir =~ /Transcripts/i){ 
          push (@mydirs, $thispath);
          next;
          }
       elsif ($adir =~ /Admin/){
          opendir(ADMIN, $thispath) or die "can't look through $thispath\n";
          undef $found;
          while ($afile = readdir(ADMIN)){
            if ($file =~ /^\./){ next;}
            if ($file =~ /^$coll\.xml$/){
              $found = 1;
              }
            elsif ($file =~ /^$coll\.\d{1,2}\.xml$/){
              $found = 1;
              }
            }
          if (!$found){ push (@missing, "$coll.xml   missing from $parent\n");}
          close(ADMIN);
          }
       elsif ($adir =~ /Metadata/){
          opendir(MD, $thispath) or die "can't look through $thispath\n";
          undef $found;
          while ($afile = readdir(MD)){
            if ($file =~ /^\./){ next;}
            if ($file =~ /^$coll\.txt$/){
              $found = 1;
              }
            elsif ($file =~ /^$coll\.\d{1,2}\.txt$/){
              $found = 1;
              }
            }
          if (!$found){ push (@missing, "$coll.txt   missing from $parent\n");}
          close(MD);
          }
  
       
       elsif ($adir =~ /$coll/){  # passes first test, it matches collname
        
         if ( -d $thispath){ 
             if (!( $adir =~ /^[a-z]{1}\d{4}\_\d{7}\_\d{7}(\_\d{4}(\_\d{3})?)?$/)){
                push (@badform, $adir);
                }
            push (@mydirs, $thispath); # collect subdirectories for further investigation    
            }
         else{  # not a directory;  must be a file. Does it match its parent directory?
            
            if ((!($parent =~ /Scans/i))  && (!($adir =~  /$parent/))){ 
                push (@wrongdir, $adir."   ".$short);
                }
            if (!( $adir =~ /^[a-z]{1}\d{4}\_\d{7}\_\d{7}(\_\d{4}(\_\d{3})?)?\.[a-z]{3}$/)){  #note this breaks on jp2, mp3
                push (@badform, $adir);
                }
        # pull off the last set of numbers before the extension   
            elsif ($parent =~ /Scans/i  || (!$parent)){ next;}  # don't check non-compound objects for sequence
            elsif ($adir =~ /.*\_(\d{3,7})\./ && (!($adir =~ /\.txt/))) {  # check for bad numbering  ONLY on files named correctly
               $mynum = $1 + 0;  # get rid of leading zeros
               if ($mynum != $count){ 
                   push (@badcount, $adir."     ".$count);
                   }
               $count ++;
               }
             }
           }
       else { push (@wrongdir, $adir."   ".$short);}
       }
       
     close(DIR);
     }



   if (@badform || @badcount || @wrongdir || @missing ){
       print OUT "\nTROUBLE: $topdir \n";
       print OUT "-------------------------------------------\n";
       }
   else{
       print OUT "All is GREAT in $topdir!  :-) \n";
       }
  
  
   if (@missing){
        print OUT "\nThe following files are missing or badly named \n";
        foreach (@missing){ print OUT "    $_\n";}
        }

   if (@wrongdir){
        print OUT "\nThe following files or directories do NOT reflect the name of their parent directory\n";
        print OUT "Are they in the right place?  Please check:\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 filenames were expected to be the sequence number\n";
        print OUT "that follows them:\n";
        foreach (@badcount){ print OUT "    $_\n";}
        }

     undef @badform;
     undef @wrongdir;
     undef @badcount;
     undef @missing
     }   # do the next top directory 
close OUT;


print "Good bye!\n";
exit;



sub timestamp{
#following for Windows
#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, $isdst) = gmtime();
##$year += 1900;
$mon ++;
if ($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";
}