#!/usr/bin/perl
use DBI;
use File::Copy;

# linkContent

# links content into EAD, using PURLs.
# this version for Cabaniss, to run off pre-generated jpegs on libcontent1
# copies this EAD into backup file first, then overwrites it with altered version.
# in this situation, all jpegs are in one directory, and there are 3 versions.  We'll
# select the one ending in _2048.jpg to use (the largest)

#  jpeg filenames reflect the EAD id
#  this script parses box number, folder number
#  pulls out filename ids of items (not their subpages)
#  turns that into a link: http://acumen.lib.ua.edu/filenameID
#  exchanges it for a PURL
# locates the correct EAD in the folder EADs
# reads it in, locates the correct box number, folder number
# and looks through ids of links already added to place this one in order

# prints out altered EAD in place of previous version, 
# with component links.

#  jody DeRidder, 12/10/09


# form of tag insert, per Shawn Averkamp:
#<c id="ref123458" level="item">
#	<did>
#		<unittitle>[whatever we want to call the item]</unittitle>
#		<dao id="u0003_0000252_00item3" ns2:title=" u0003_0000252_00item3" ns2:href="http://item.purl/3" ns2:actuate="onRequest" ns2:show="new"/>
#	</did>
#</c>
##!!! Altered 12.15.09 by Shawn to add <dao> AFTER item <did />

#<c id="ref8" level="item">
#<did>
#<unittitle>Francis William Kellogg Letter</unittitle>
#<container type="Box">Sc0004</container>
#<container type="Folder">2027.01</container>
#</did>
#<dao id="u0003_0002027_0000001" ns2:actuate="onRequest" ns2:show="new" ns2:title="u0003_0002027_0000001" ns2:role="Image-Service" ns2:href="http://purl.lib.ua.edu/493"/>
#</c>

#   and per Donnelly Lancaster Walton, the [whatever we want to call the item] should be:
#  Item #   -- reflecting the sequence, starting at one, for each folder.
# assume that the files in each folder are complete and in sequence.

# ref numbers need to be unique.  Given that item  numbers are unique, we will use those

# note that this assumes breakdown into series can occur at the folder level, but not below;
# boxes 21 and 26 of Cabaniss contain content in 2 different series
#  earlier version by  jody DeRidder, 11/16/09

$eadDir = "../EAD/";
$backups = "../backups/";
$content = "../jpegs/";
$coll = "u0003_0000252";
$ead = $coll.".ead.xml";

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


# script begins by backing up current EAD with timestamp
# reading it in and extracting out IDs of content already linked.

&timestamp;
$backup = $ead.$timestamp;
$output = ">../output/Linking_README_$timestamp";
open (OUTPUT, $output) or die "cannot write output errors\n";


$old = $eadDir.$ead;
$new = $backups.$backup;
copy ($old, $new) or die "could not make a backup copy of $ead!\n";

open (IN, $old) or die "can't read $ead\n";
while ($line = <IN>){
  if ($line =~ /<dao  *id="([^"]+)"/){
     push (@linked, $1);
     }
  push (@thisEad, $line);
  }
close(IN);

# before I unlink this and open it anew, let's verify that we need to do something to it, shall we?
# picking up the item numbers from the jpegs next, and comparing them to the list I just put in 
#  the @linked array.

opendir(JPEGS, $content) or die "can't look through $content\n";
while ($file = readdir(JPEGS)){
    if ($file =~ /^\./){
      next;  # skip dot files
      }
    elsif ($file =~ /^($coll\_\d{7}).*?\_2048\.jpg$/){   # the largest;  be aware there may be
       $item = $1;
   #    print "$item for $file\n";
       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;
           }
         }
       if (! $found){
         foreach $done (@linked){
           if ($done eq $item){ $found = 1;}   # now check to see if I linked it already
           }
         if (!$found){ push (@doThese, $item);}  # if not, put it on my todo list
         }
      }
    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";}


# delete old copy of EAD and reopen it.
unlink $old;
$ead = ">$old";
open (OUT, $ead) or die "can't open $ead\n";

# get the last purlnum from the database, and lock the tables...
$dbh = DBI->connect("DBI:mysql:$database:$hostname:$port",
      $user, $password) or die "can't connect to database: ",$DBI::errstr,"\n";

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

$sth = $dbh->prepare(" lock table lookup write")
    or die "Can't lock dbase tables!! : ",$dbh->errstr(),"\n";
$sth->execute()
    or die "Can't execute SQL statement: ", $sth->errstr(),"\n";

$sth = $dbh->prepare(" select max(purlnum) from lookup")
    or print "Can't get a count! Kill me and call for help!! : ",$dbh->errstr(),"\n";
$sth->execute()
    or die "Can't execute SQL statement: ", $sth->errstr(),"\n";
$lastone = $sth->fetchrow_array();
     warn "Problem in fetchrow_array(): ",$sth->errstr(),"\n" if $sth->err();
$lastone ++;

@mylist = sort by_number (@doThese);  

foreach $item (@mylist){

    if ($item =~ /^[a-z]{1}\d{4}\_\d{7}\_(\d{2})(\d{2})(\d{3})$/){
       $box= $1;
       $folder= $2;
       $thisItem = $3;
       $location{$item} = $box.'_'.$folder;

# this puts the items in order within folder within boxes

       push (@{$whatwhere{$box}{$folder}}, $item); 
     #  print "$box, $folder for item $item\n";
       }    
    else{ push(@errors, "ERROR: Filename format is incorrect: -->$item<--\n");}
    }

@mybox = sort by_number( keys (%whatwhere));
print " here's my boxes:\n";
foreach (@mybox){ print "box $_\n";}

# here's what a box looks like.  This is box 1, folder 1.  Folders  for a single box
#    may be in different places in the EAD:
#                     <c id="ref22" level="file">
#                        <did>
#                            <unittitle>Incoming, A - B. Cabaniss</unittitle>
#                            <container type="Box">252.001</container>
#                            <container type="Folder">1</container>
#                        </did>
#                    </c>

# all items need to be entered  after the </did> and prior to the closing "c" for this folder.

# since folders of a single box may be in multiple places, we need to run through the EAD
# once per box.
@changed = @thisEad;
foreach $abox (@mybox){
     $lookfor = sprintf("%03d", $abox);  # leftpad to 3 places.

     @myfolders = sort by_number (keys(%{$whatwhere{$abox}}));
     foreach $afolder (@myfolders){
       undef @notMe;    # to contain list of items already in EAD
   #    print "$abox contains folder $afolder\n";
       undef @myItems;
       @myItems = @{$whatwhere{$abox}{$afolder}};
       $afolder += 0;  # remove left padding

    #   foreach (@myItems){ print "Item $_\n";}
       undef $gotIt;
       undef $foundBox;    # exists when we find that box
       undef $foundFolder;  # exists when we find one of the needed folders
       undef $foundDid;   # exists when we've seen the close of the folder did
       undef $foundC;   # need to add new values BEFORE this;  finding this undefs the others, go to next folder
       undef @thisRound;
       foreach $line (@changed){
          if ((! $foundBox) && $line =~ /<container type="Box"> *\d*\.$lookfor *<\/container>/){
            $foundBox = 1;
           # print "found box $lookfor\n";
            push (@thisRound, $line);
            }
          elsif($foundBox && (!$foundFolder) && (!$foundC)){
            if ($line =~ /<\/c>/){   # this folder is listed somewhere else, or does not exist.
                 # need to keep looking through the EAD.
              undef $foundBox;
              }
           elsif ($line =~ /<container type="Folder"> *$afolder *<\/container>/){
              $foundFolder = 1;
           #   print "found folder $afolder\n";
              $gotIt = 1;
              }
            push (@thisRound, $line);
            undef $foundC;
            }
          elsif($foundBox && $foundFolder && (!$foundC)){
         #    print "looking for c.\n";
             if ($line =~ /<c id="ref[^"]*" *level="item">/){
                $closeChild = 1;  # close the kid first.
                }
             elsif ($closeChild && $line =~ /dao id="([^"]*)"/){
               $itemThere = $1;
               undef $found;
               foreach $i (@myItems){  
                 if ($i eq $itemThere){ $found = 1;}
                 }
               if ($found){
                  push (@errors, "CHECK:  $itemThere is already in the EAD\n");
                  push (@notMe, $itemThere);
                  }
               }
             elsif($line =~ /<\/c>/ && $closeChild){
                undef $closeChild;
                }

             elsif ($line =~ /<\/c>/){
                $foundC = 1;
#                print "found the end of that c -- inserting items\n";
                # must insert items before this line
#<c id="ref123458" level="item">
#       <did>
#               <unittitle>[whatever we want to call the item]</unittitle>
#               <dao id="u0003_0000252_00item3" ns2:title=" u0003_0000252_00item3" ns2:href="http://item.purl/3" ns2:actuate="onRequest" ns2:show="new"/>
#       </did>
#</c>

                # $count = 1;
                 foreach $item (@myItems){
                   undef $found;

                   # first check to see if it's already in there
                   foreach $i (@notMe){
                      if ($i eq $item){ $found = 1;}
                      }
                   if ($found){next;}  # skip this one, it's in there
           
# instead of using count to give item number, let's use the last 3 digits of the item +0
#  an item might not be digitized on purpose.  
                   ($myname = $item) =~ s,.*?(\d{3})$,\1,; 
                   $myname += 0;                  
               #    print "$item is Item $myname\n";
                   undef $mypurlnum;
                   $ref = "ref".$item;

# 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, "CHECK:  $id2009 already has a PURL:\n$mypurlnum  $myrealurl\n");
                     }
                 else{    # insert this baby
                      $mypurlnum = $lastone;
                      $purlnum = $dbh->quote($lastone);
                      $lastone ++;
                      $url = "http://acumen.lib.ua.edu/".$item;
                      $myurl = $dbh->quote($url);
                      $sth = $dbh->prepare("insert into lookup (dnum, id_2009, purlnum, realurl, datestamp, history) values (NULL,$id2009,$purlnum,$myurl,NULL,NULL)")
                         or die "can't prepare insert for $item: ",$dbh->errstr(),"\n";
                      $sth->execute()
                         or die "Can't insert into lookup $id2009 : ", $sth->errstr(),"\n";

                      $sth->finish();
                      }
               #    print "$item  purlnum is $mypurlnum\n";

                   push (@thisRound, "                        <c id=\"$ref\" level=\"item\">\n");
                   push (@thisRound, "                           <did>\n");
                   push (@thisRound, "                             <unittitle>Item $myname</unittitle>\n"); 
                   push (@thisRound, "                           </did>\n");
                   push (@thisRound, "                           <dao id=\"$item\" ns2:title=\"$item\" ns2:href=\"http://purl.lib.ua.edu/".$mypurlnum."\" ns2:actuate=\"onRequest\" ns2:show=\"new\"/>\n");
                   push (@thisRound, "                         </c>\n");
                  # $count ++;
                   }
               undef $foundBox;
               undef $foundFolder;
               
               }
             push (@thisRound, $line);

             }
           else{  push (@thisRound, $line);} 
           } 
        @changed = @thisRound;   # keep altering EAD till everything found and linked, then write it out
        if (!$gotIt){ push(@errors,  "ERROR!!!  $abox  $afolder  is NOT IN THE EAD!!  Items NOT linked!!!\n");} 
        }
     } 
foreach (@changed){ print OUT $_;}
close(OUT);

$sth = $dbh->prepare(" unlock tables")
    or die "Can't unlock dbase tables!! : ",$dbh->errstr(),"\n";
$sth->execute()
    or die "Can't execute SQL statement: ", $sth->errstr(),"\n";

$dbh->disconnect();

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

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

   ($sec, $min, $hour, $mday, $mon, $year, $wday, $yday, $isdst) = localtime();  #gmtime();
   $mon ++;
   if ($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";
   }