<?php
/*
 * This script creates a Turtle file containing rdfs:seeAlso mappings between MeSH (as represented in the Neurocommons KB) and DBpedia. The mapping is based on data in the OBO ontologies.
 *
 * Note that this script requires the ARC2 library. You need to download this library separatly.
 * 
 * This work is in the public domain. 
 * Matthias Samwald, 2009. http://samwald.info
 */

/* ARC2 library must be present! */ 
include_once('arc/ARC2.php');

// read mapping from MeSH labels to MeSH URIs into array
$mesh_uri_label_mapping_file = fopen("mesh_uri_label_mapping.txt", "r");
$mesh_uri_label_mapping = Array(); // mapping between MeSH labels to MeSH URIs
while (($data = fgetcsv ($mesh_uri_label_mapping_file, 1000, "\t")) !== FALSE ) {
	$mesh_uri_label_mapping[$data[1]] = $data[0];
  }
fclose($mesh_uri_label_mapping_file);


$output_file_handle = fopen("dbpedia_mesh_mapping.ttl", "w");


// write namespace prefixes
fputs($output_file_handle, "@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> . \n\n");

$config = array(
  'remote_store_endpoint' => 'http://lod.openlinksw.com/sparql',
);

$store = ARC2::getRemoteStore($config);

$offset = 0;

$q = 'SELECT * WHERE {?dbpedia_uri <http://dbpedia.org/ontology/meshid> ?mesh_id . } LIMIT 200 OFFSET ';

while ($rows = $store->query($q . $offset, 'rows')) {
  print("\n offset: " . $offset);
  foreach ($rows as $row) {
  	fputs($output_file_handle, '<' . $row['dbpedia_uri'] . '> rdfs:seeAlso <http://purl.org/commons/record/mesh/' . $row['mesh_id'] . "> . \n");
  }  	
  $offset = $offset + 200;
  print("\n offset: " . $offset);
}

fputs($output_file_handle, "\n\n");

$offset = 0;

$q = 'SELECT * WHERE {?dbpedia_uri <http://dbpedia.org/property/meshname> ?mesh_name . } LIMIT 200 OFFSET ';

while ($rows = $store->query($q . $offset, 'rows')) {
  print("\n offset: " . $offset);	
  foreach ($rows as $row) {
  	$dbpedia_uri = $row['dbpedia_uri'];
  	$dbpedia_uri = str_replace("/section1/Chembox_Identifiers", "", $dbpedia_uri);
  	
  	$mesh_name = $row['mesh_name'];  	
  	$mesh_name = str_replace("+", " ", $mesh_name);
  	$mesh_name = str_replace(" }}", "", $mesh_name);
  	
  	$mesh_uri = $mesh_uri_label_mapping[$mesh_name];
  	if ($mesh_uri == "" ) { continue; }
  	  		
  	fputs($output_file_handle, '<' . $dbpedia_uri . '> rdfs:seeAlso <' . $mesh_uri . "> . \n");
  }  	
  $offset = $offset + 200;  
}

fclose($output_file_handle);

print "\n -- done -- ";
?>