<?php

# ------------------------------------------------------------------------- #
# Scriptname:		cdmtab2dspace-dist 
#
# Author:		John Millard <millarj@muohio.edu>
# Purpose: 		Format a tab delimited export from CONTENTdm into a 
# 			dSpace Bulk Ingest package
#
# Requirements:: 
#			- PHP 5.* CLI with fopen enabled
# 			- Admin access to the CONTENTdm collection 
# 			- filesystem access to the digital objects
#
# Caveat:		This is pretty ugly code. Apologies in advance.
# 			
# ------------------------------------------------------------------------- #

# --- Initial Configuration Options:  

# php.ini override to attempt to detect line ending type
# Comment it out if you don't want/need it.

   ini_set('auto_detect_line_endings', 1);

# Test mode (on/off):  on: Prints out what WOULD happen.  Does not execute
#      	    	       off: Proceeds with creating and populating the package
   $test_mode = "on";


# Location of physical items on the local filesystem
   $path_to_items = "/path-to-items/";

# Location of Destination project - Created if doesn't exist
   $project_home = "/project_home/";

# The shortname used by Contentdm or any other unique project label
   $collection_shortname = "uw";

# Name and location of mapped tab-delimited data file
   $tab_metadata_file = "sample-export-modified.txt";

# An array of patterns to exclude from output if lines match.  
# Useful for excluding compound object records if flattening the object 
  $exclude_filters = array
  (
  "Side *1",
  "Side *2"
  );

# End Configuration Options

# Destination on the local filesytem for the dSpace Bulk Ingest package
  $project_base_dir = "$project_home$collection_shortname";

# Create initial directories -- Check for existence first

  if (!file_exists($project_base_dir)&&$test_mode == "off") {mkdir($project_base_dir, 0700, TRUE);}

$filter = implode("|", $exclude_filters);
$input_file = file($tab_metadata_file);
$drc_labels = split("\t", $input_file[0]);
$cdm_labels = split("\n", $input_file[1]);

# lop the first two lines from the array of lines to get dspace --> cdm mapping

  $input_file = array_slice($input_file, 2);

  foreach ($input_file as $unf_line)
  {
    if(!preg_match("/$filter/",$unf_line)) { $metadata[] = $unf_line;}
  }

  foreach ($metadata as $line)
  {

  $fields = split("\t", $line);

# Combine drc_labels and fields arrays using drc_labels as keys and fields as values
  $drc_metadata = array_combine($drc_labels, $fields);


# output xml declaration

  $dc_record = "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\n";

  foreach($drc_metadata as $key => $value) 
  {

  if(!preg_match("/cdmfile|subject|skip/",$key)) 
  {
	if (ereg("\.", $key)) 
	{ 
	$qualifiers = split("\.", $key);
	$dc_record = $dc_record . "<dcvalue element=\"$qualifiers[0]\" qualifier=\"$qualifiers[1]\">$value</dcvalue>\n";
	} else 
	{
	$dc_record = $dc_record . "<dcvalue element=\"$key\" qualifier=\"none\">$value</dcvalue>\n";
	}

  }


# Process Subjects - This get's ugly

  if (ereg("subject", $key))
   {
	$subjects = split(";", $value);

	if (ereg("\.", $key)) 
	{ 
	$qualifiers = split("\.", $key);
	foreach ($subjects as $sub)
		{ 	
		if ($sub != "") {$dc_record = $dc_record . "<dcvalue element=\"$qualifiers[0]\" qualifier=\"$qualifiers[1]\">$sub</dcvalue>\n";}
		}
	} else 
	{
	foreach ($subjects as $sub)
		{ 	
		if ($sub != "") {$dc_record = $dc_record . "<dcvalue element=\"$key\" qualifier=\"none\">$sub</dcvalue>\n";}
		}
	}
   }
}

# output closing
$dc_record = $dc_record . "</dublin_core>\n";

$counter++;


$item_path = "$project_base_dir/$counter";

# Call assembly functions
  make_item_folder($item_path);
  output_dcrecord($dc_record);
  move_bitstreams($item_path);
  if ($test_mode == "on") { print " --- End of Record ---\n";}

}

# Begin Additional Functions

function output_dcrecord($dc_record) 
{
  global $dc_record, $item_path, $test_mode;
  if ($test_mode == "off")
  {
   $metadata_handle = fopen("$item_path/dublin_core.xml", "w+");
   fwrite($metadata_handle, $dc_record);
   fclose($metadata_handle);

   } else {
   print "TEST dc xml: $dc_record\n";
   }

}

function make_item_folder($item_path)
{
global $item_path, $test_mode;
  if (!file_exists($item_path)&&$test_mode == "off") {mkdir($item_path, 0700);} else
  {
  print "TEST setting item path: Item path would be: $item_path\n";
  }
}

function move_bitstreams($item_path)
{
# Customized code to locate and copy bitstreams into bundle
# Additionally, needs to write out the content manifest based on files moved
# Generic example follows:

global $item_path, $drc_metadata, $path_to_items, $test_mode;

$filename = $drc_metadata["identifier.other"];
$manifest = $manifest . "$filename\tbundle:ORIGINAL\n";

if ($test_mode == "off") 
{
copy("$path_to_items$filename", "$item_path/$filename");
$contents_handle = fopen("$item_path/contents", "w+");
fwrite($contents_handle, $manifest);
fclose($contents_handle);
}
else {
print "TEST copy bitstreams: cp $path_to_items$filename $item_path/$filename\n";
print "TEST manifest: $manifest\n";
}

}


?>