#!/usr/bin/perl
use DBI;

# generateMods.pl

# this version to be run from jeremiah's home directory, on Cabaniss
# working over jpegs all in the same directory, instead of tiffs

# checks first for MODS in web directory.  
# if there, will not create MODS for that item

#  expects filenames that reflect the EAD id
#  this script parses box number, folder number
#  pulls out filename ids of items 
# looks for number of pages belonging to this item

# looks up PURL in database
# if not there, outputs error

# takes MODS template, and substitutes in values for this item
#  Item number (for title), box, folder, extent, PURL
#  outputs a MODS for this item into MODS directory, named for the 
#  item ID

# puts copies for archiving into deposits directory

# jody DeRidder, 12/11/09

$modsDir = "../MODS/";
$content = "../jpegs/";
$template = "./MODStemplate.xml";


$timestamp = &gettime;
$output = "../output/MODS_README_$timestamp";
open (OUT, ">".$output) or die "can't open $output\n";

# values to replace in template include:
# TITLE_HERE, EXTENT_HERE, PURL_HERE, FILEID_HERE, BOX_HERE, FOLDER_HERE
# and of course, the filename.

$coll = "u0003_0000252";
$modsCopies = "/srv/deposits/Cabaniss/u0003_0000252_Cabaniss/Metadata/";
$webdirs = "/srv/www/htdocs/content/u0003/0000252/";

push (@mydirs, $webdirs);
foreach $dir (@mydirs){
  opendir(DIR, $dir) or die "can't open $dir\n";
  while ($file = readdir(DIR)){
    if ($file =~ /^\./){ next;} # skip dot files
      $path = $dir.$file;
      if (-d $path){ push (@mydirs, $path."/");}
      elsif ($file =~ /^(.*?)\.mods\.xml/){
        push (@gotThese, $1);
        }
      }
   close(DIR);
   }

# now we have all ids for existing MODS in @gotThese

# put database login stuff here
$hostname = "localhost";
$port = "3306";
$user = "addData";
$password = "moreStuff";
$database = "InfoTrack";

$dbh = DBI->connect("DBI:mysql:$database:$hostname:$port",
      $user, $password) or die "can't connect to database: ",$DBI::errstr,"\n";

$h->{PrintError} = 1;
$h->{RaiseError} = 1;

# first, get the template
open(TEMP, $template) or die "can't read in $template\n";
while ($line = <TEMP>){
  push (@template, $line);
  }
close (TEMP);


opendir(JPEGS, $content) or die "can't look through $content\n";
while ($file = readdir(JPEGS)){
    if ($file =~ /^\./){
      next;  # skip dot files
      }
# this does NOT process subpages!
 #   print "looking at $file\n";
    if ($file =~ /^$coll\_(\d{2})(\d{2})(\d{3})(\_(\d{4}))?\_2048\.jpg$/){   # the largest;  be aware there may be
       $item = $coll."_".$1.$2.$3;
       $box = $1;
       $folder = $2;
       $item_sequence = $3 + 0; # gets rid of padding zeros

   #    print "found $item\n";
       if ($4){
          $thisPage = $5 + 0;  #remove padding zeros
          }
       else{ undef $thisPage;}
       undef $found;
       foreach $todo (@doThese){  # first check to see if I picked up one of this item's images
         if ($todo eq $item){     # already.  This list should be shorter soon than the other
           $found = 1;
           $ItemExtent{$item} += 1;  # add a page
   #        print "   already on my list.\n";
           }
         }
       if (! $found){
         foreach $done (@gotThese){
           if ($done eq $item){ $found = 1;} #  print "   Already have a MODS\n";}   # now check to see if I made the MODS already
           }
         if (!$found){
            push (@doThese, $item);  # if not, put it on my todo list
          #  print "Need to process $item\n";
            $ItemExtent{$item} = 1;
            $ItemFolder{$item} = $folder;
            $ItemBox{$item} = $box;
            $ItemTitle{$item} = "Item $item_sequence";

            }
         }
      }
    elsif (!( $file =~ /\_512.jpg$/ || $file =~ /\_128.jpg$/ || $file =~ /thumbs\.db/i)){
       push(@errors, "ERROR! Check filename. DID NOT PROCESS: $file\n");
       }
    }
close(JPEGS);


$num = scalar (@doThese);
if ($num < 1){ print "Did not find any new items to link!!  Exiting.\n";  exit;}
else{ print "Found $num new items to link.  Processing now\n";}


$subtract = 0;
   
$all = (scalar @doThese);
while (($item, $extent) = each (%ItemExtent)){
   $digFiles += $extent;
   }

print OUT "\nFound $all items for new MODS, consisting of $digFiles digital files\n";


@allitems = sort by_number (keys (%ItemTitle));
foreach $item (@allitems){

           
   undef $mypurlnum;

# fetch a purl here
# check first to see if we have one yet for this item
     $id2009 = $dbh->quote($item);
     $sth = $dbh->prepare("select purlnum, realurl from lookup where id_2009 = $id2009")
           or die "can't look for $id2009 in database: ",$dbh->errstr(),"\n";
     $sth->execute()
        or die "Can't look for $id2009 in lookup : ", $sth->errstr(),"\n";

     ($mypurlnum, $myrealurl) = $sth->fetchrow_array();
       warn "Problem in fetchrow_array(): ",$sth->errstr(),"\n" if $sth->err();
      $sth->finish();
      if (! $mypurlnum){
           push (@errors, "ERROR:  $id2009  has NO PURL! Did NOT make MODS\n");
           }
      else{  
           $url = "http://purl.lib.ua.edu/".$mypurlnum;

   #        print "$item:  ".$ItemTitle{$item}."    Extent:  ".$ItemExtent{$item}."\n";
    #       print "        Box  ".$ItemBox{$item}."  Folder ".$ItemFolder{$item}."\n";
     #      print "        PURL:  $url\n";


# values to replace in template include:
# TITLE_HERE, EXTENT_HERE, PURL_HERE, FILEID_HERE, BOX_HERE, FOLDER_HERE
# and of course, the filename.
            $mymods = $modsDir.$item.".mods.xml";
            open (MYMODS, ">".$mymods) or die "can't write to $mymods\n";
            foreach $line (@template){
              if ($line =~ /^(.*?)TITLE_HERE(.*)/){
                 print MYMODS $1.$ItemTitle{$item}.$2."\n";              
                 }
              elsif ($line =~ /^(.*?)EXTENT_HERE(.*)/){           
                 print MYMODS $1.$ItemExtent{$item}." p.".$2."\n"; 
                 }
              elsif ($line =~ /^(.*?)PURL_HERE(.*)/){     
                 print MYMODS $1.$url.$2."\n";
                 }
              elsif ($line =~ /^(.*?)FILEID_HERE(.*)/){
                 print MYMODS $1.$item.$2."\n";
                 }

              elsif ($line =~ /^(.*?)BOX_HERE(.*)/){     
                 print MYMODS $1.$ItemBox{$item}.$2."\n";
                 }
              elsif ($line =~ /^(.*?)FOLDER_HERE(.*)/){
                 print MYMODS $1.$ItemFolder{$item}.$2."\n";
                 }
              elsif ($line =~ /^(.*?)DATE_HERE(.*)/){
                 print MYMODS $1.$date.$2."\n";
                 }

              elsif ($line =~ /^(.*? PRELIMINARY METADATA GENERATED BY SCRIPT )(.*)/){
                 print MYMODS $1.$timestamp."  $2\n";
                 }
              else{  print MYMODS $line;}
              }
            close (MYMODS);

            $copy = $modsCopies.$item.".mods.xml";
            copy ($mymods, $copy) or push(@errors, "WARNING: Could not copy $mymods to $copy\n";);
            }  
         }      


$dbh->disconnect();

if (@errors){
   print "All done. Check the README file in the output directory for errors, please!\n";
   foreach (@errors){ print OUT$_;}
   sleep(3);
   close(OUT);
   }
else{
  print "All done!  Please check MODS directory for results.  Thank you!\n";
  sleep(3);
  }

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

sub gettime{
# $date = `date`;
# $time = `time`;

 ($sec, $min, $hour, $mday, $mon, $year, $wday, $yday, $isdst) = localtime();  #gmtime();  #$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;
 $date = "$year-$mon-$mday";
 return  "$year-$mon-$mday\T$hour:$min:$sec\Z";
}