<?php

/*
 * This script creates a Turtle file containing rdfs:seeAlso mappings between OBO ontologies that are hosted on the hcls.deri.org/sparql SPARQL endpoint 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');

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

/*
 *  Function for extracting IDs that can be used to build DBpedia URIs
 */

function extract_id($string) {
	
	if (strpos($string, "en.wikipedia.org/wiki/") !== false) {
		$string_chunk_array = explode("en.wikipedia.org/wiki/", $string);
  		$string = $string_chunk_array[1];
	}
  	elseif (strpos($string, "Wikipedia#Wikipedia_") !== false) {
		$string_chunk_array = explode("Wikipedia#Wikipedia_", $string);
  		$string = $string_chunk_array[1];
	}
	else {
		return "";
	}
		
	// return empty string if the reference is non-sensical (i.e., just a general reference to Wikipedia itself)
	if (strpos($string, "The_Free_Encyclopedia") !== false) {
		return "";
	}
	if (strpos($string, "Wikipedia") !== false) {
		return "";
	}
	if (strpos($string, "wikipedia") !== false) {
		return "";
	}

	  	
	// if string contains hashes or characters that do not belong to the URI, lop off the trailing part of the string
	$position_of_first_bad_character = strcspn($string, " #<>[]");
	$string = substr($string, 0, $position_of_first_bad_character);
	
	// if string ends with ) and there is no prior ( in string, then the ) does not belong to the id
	if (((strpos($string, ")") !== false)) && (strpos($string, "(") !== true)) {
		$position_of_first_bad_character = strcspn($string, ")");
		$string = substr($string, 0, $position_of_first_bad_character);
	}
	
	return $string;
}

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

// ------- hasDbXref/hasURI query ------- 

$config = array(
  'remote_store_endpoint' => 'http://sparql.obo.neurocommons.org/sparql',
);

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

$q = 
'PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX obo: <http://www.geneontology.org/formats/oboInOwl#>

SELECT DISTINCT ?entity ?wikipedia_uri 
WHERE {
?entity obo:hasDefinition ?definition_bnode .
?definition_bnode obo:hasDbXref ?dbxref_bnode .
?dbxref_bnode obo:hasURI ?wikipedia_uri .
FILTER regex(str(?wikipedia_uri), "wikipedia", "i")
}
';

if ($rows = $store->query($q, 'rows')) {
  foreach ($rows as $row) {
  	
  	$id = extract_id($row['wikipedia_uri']);
  	
  	if ($id != "") {
    	fputs($output_file_handle, '<' . $row['entity'] . '> rdfs:seeAlso <http://dbpedia.org/resource/' . urldecode($id) . "> . \n");
  	}
  }
}

// ------- OBI query ------- 

$config = array(
  'remote_store_endpoint' => 'http://sparql.obo.neurocommons.org/sparql',
);

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

$q ='
PREFIX owl:<http://www.w3.org/2002/07/owl#>
PREFIX obi:<http://purl.obofoundry.org/obo/>
PREFIX rdfs:<http://www.w3.org/2000/01/rdf-schema#>

SELECT DISTINCT ?entity ?wikipedia_uri
WHERE {
?entity obi:IAO_0000119 ?wikipedia_uri.
FILTER regex(str(?wikipedia_uri), "en.wikipedia.org/wiki/", "i")
}
';

if ($rows = $store->query($q, 'rows')) {
  foreach ($rows as $row) {

  	$id = extract_id($row['wikipedia_uri']);
  	
  	if ($id != "") {
    	fputs($output_file_handle, '<' . $row['entity'] . '> rdfs:seeAlso <http://dbpedia.org/resource/' . urldecode($id) . "> . \n");
  	}
  }
}

fclose($output_file_handle);

print " script succesfully executed. ";

?>