Skip to content

Commit 75cb6ac

Browse files
alisoncallahanalisoncallahan
committed
Started updating BioModels parser to release 3
1 parent 1820284 commit 75cb6ac

File tree

1 file changed

+84
-70
lines changed

1 file changed

+84
-70
lines changed

‎biomodels/biomodels.php

+84-70
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
<?php
22
/**
3-
Copyright (C) 2012 Michel Dumontier
3+
Copyright (C) 2013 Michel Dumontier, Alison Callahan
44
55
Permission is hereby granted, free of charge, to any person obtaining a copy of
66
this software and associated documentation files (the "Software"), to deal in
@@ -20,50 +20,36 @@
2020
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
2121
SOFTWARE.
2222
*/
23-
24-
require('../../php-lib/biopax2bio2rdf.php');
25-
2623
/**
2724
* BioModels RDFizer
28-
* @version 1.0
25+
* @version 2.0
2926
* @author Michel Dumontier
27+
* @author Alison Callahan
3028
* @description http://www.ebi.ac.uk/biomodels-main/
3129
*/
32-
class BiomodelsParser extends RDFFactory
30+
31+
require_once(__DIR__.'/../../php-lib/biopax2bio2rdf.php');
32+
33+
class BiomodelsParser extends Bio2RDFizer
3334
{
34-
private $version = null;
3535

3636
function __construct($argv) {
37-
parent::__construct();
38-
$this->SetDefaultNamespace("biomodels");
37+
parent::__construct($argv, "biomodels");
3938

4039
// set and print application parameters
41-
$this->AddParameter('files',true,null,'all|curated|biomodel#|start#-end#','entries to process: comma-separated list or hyphen-separated range');
42-
$this->AddParameter('indir',false,null,'/data/download/'.$this->GetNamespace().'/','directory to download into and parse from');
43-
$this->AddParameter('outdir',false,null,'/data/rdf/'.$this->GetNamespace().'/','directory to place rdfized files');
44-
$this->AddParameter('graph_uri',false,null,null,'provide the graph uri to generate n-quads instead of n-triples');
45-
$this->AddParameter('gzip',false,'true|false','true','gzip the output');
46-
$this->AddParameter('download',false,'true|false','false','set true to download files');
47-
$this->AddParameter('download_url',false,null,'http://www.ebi.ac.uk/biomodels/models-main/publ/');
48-
if($this->SetParameters($argv) == FALSE) {
49-
$this->PrintParameters($argv);
50-
exit;
51-
}
52-
if($this->CreateDirectory($this->GetParameterValue('indir')) === FALSE) exit;
53-
if($this->CreateDirectory($this->GetParameterValue('outdir')) === FALSE) exit;
54-
if($this->GetParameterValue('graph_uri')) $this->SetGraphURI($this->GetParameterValue('graph_uri'));
55-
56-
return TRUE;
40+
parent::addParameter('files',true,null,'all|curated|biomodel#|start#-end#','entries to process: comma-separated list or hyphen-separated range');
41+
parent::addParameter('download_url',false,null,'http://www.ebi.ac.uk/biomodels/models-main/publ/');
42+
parent::initialize();
5743
}
5844

5945
function Run()
6046
{
6147
// directory shortcuts
62-
$ldir = $this->GetParameterValue('indir');
63-
$odir = $this->GetParameterValue('outdir');
48+
$ldir = parent::getParameterValue('indir');
49+
$odir = parent::getParameterValue('outdir');
6450

6551
// get the work specified
66-
$list = trim($this->GetParameterValue('files'));
52+
$list = trim(parent::getParameterValue('files'));
6753
if($list == 'all') {
6854
// call the getAllModelsId webservice
6955
$file = $ldir."all_models.json";
@@ -108,29 +94,33 @@ function Run()
10894
}
10995
}
11096
}
97+
98+
$graph_uri = parent::getGraphURI();
99+
if(parent::getParameterValue('dataset_graph') == true) parent::setGraphURI(parent::getDatasetURI());
111100

112101
// set the write file
113-
$outfile = 'biomodels.nt'; $gz=false;
114-
if($this->GetParameterValue('graph_uri')) {$outfile = 'biomodels.nq';}
115-
if($this->GetParameterValue('gzip')) {
116-
$outfile .= '.gz';
102+
$suffix = parent::getParameterValue('output_format');
103+
$outfile = 'biomodels'.'.'.$suffix;
104+
$gz=false;
105+
106+
if(strstr(parent::getParameterValue('output_format'), "gz")) {
117107
$gz = true;
118108
}
119-
$bio2rdf_download_files[] = $this->GetBio2RDFDownloadURL($this->GetNamespace()).$outfile;
120109

121-
$this->SetWriteFile($odir.$outfile, $gz);
110+
$dataset_description = '';
111+
parent::setWriteFile($odir.$outfile, $gz);
122112

123113
// iterate over the entries
124114
$i = 0;
125115
$total = count($entries);
126116
foreach($entries AS $id) {
127117
echo "processing ".(++$i)." of $total - biomodel# ".$id;
128118
$download_file = $ldir.$id.".owl.gz";
119+
$url = parent::getParameterValue('download_url')."$id/$id-biopax3.owl";
129120
// download if the file doesn't exist or we are told to
130121
if(!file_exists($download_file) || $this->GetParameterValue('download') == 'true') {
131122
// download
132123
echo " - downloading";
133-
$url = $this->GetParameterValue('download_url')."$id/$id-biopax3.owl";
134124
$buf = file_get_contents($url);
135125
if(strlen($buf) != 0) {
136126
file_put_contents("compress.zlib://".$download_file, $buf);
@@ -139,52 +129,76 @@ function Run()
139129
}
140130

141131
// load entry, parse and write to file
142-
echo " - parsing";
132+
echo " - parsing... ";
143133
// $this->SetReadFile($download_file,true);
144134
$buf = file_get_contents("compress.zlib://".$download_file);
145-
146-
$converter = new BioPAX2Bio2RDF();
135+
136+
$converter = new BioPAX2Bio2RDF($this->getRegistry());
147137
$converter->SetBuffer($buf)
148138
->SetBioPAXVersion(3)
149139
->SetBaseNamespace("http://identifiers.org/biomodels.db/$id/")
150140
->SetBio2RDFNamespace("http://bio2rdf.org/biomodels:".$id."_")
151141
->SetDatasetURI($this->GetDatasetURI());
152-
$this->AddRDF($converter->Parse());
153-
$this->WriteRDFBufferToWriteFile();
142+
143+
$rdf = $converter->Parse();
144+
parent::addRDF($rdf);
145+
parent::writeRDFBufferToWriteFile();
146+
parent::getWriteFile()->Close();
154147

155-
echo PHP_EOL;
156-
}
157-
$this->GetWriteFile()->Close();
158-
159-
// generate the release file
160-
$this->DeleteBio2RDFReleaseFiles($odir);
161-
$desc = $this->GetBio2RDFDatasetDescription(
162-
$this->GetNamespace(),
163-
"https://github.com/bio2rdf/bio2rdf-scripts/blob/master/biomodels/biomodels.php",
164-
$bio2rdf_download_files,
165-
"http://www.ebi.ac.uk/biomodels-main/",
166-
array("use-share-modify"),
167-
null, // license
168-
$this->GetParameterValue('download_url'),
169-
$this->version
170-
);
171-
$this->SetWriteFile($odir.$this->GetBio2RDFReleaseFile($this->GetNamespace()));
172-
$this->GetWriteFile()->Write($desc);
173-
$this->GetWriteFile()->Close();
148+
echo "done!".PHP_EOL;
149+
150+
//generate dataset description
151+
echo "Generating dataset description for BioModel # $id... ";
152+
$source_file = (new DataResource($this))
153+
->setURI($url)
154+
->setTitle("EBI BioModels Database - BioModel # $id")
155+
->setRetrievedDate( date ("Y-m-d\TG:i:s\Z", filemtime($download_file)))
156+
->setFormat("rdf/xml")
157+
->setPublisher("http://www.ebi.ac.uk/")
158+
->setHomepage("http://www.ebi.ac.uk/biomodels-main/")
159+
->setRights("use-share-modify")
160+
->setLicense("http://www.ebi.ac.uk/biomodels-main/termsofuse")
161+
->setDataset("http://identifiers.org/biomodels.db/");
162+
163+
$dataset_description .= $source_file->toRDF();
164+
echo "done!".PHP_EOL;
165+
166+
}//foreach
167+
168+
echo "Generating dataset description for Bio2RDF BioModels... ";
169+
170+
$prefix = parent::getPrefix();
171+
$bVersion = parent::getParameterValue('bio2rdf_release');
172+
$date = date ("Y-m-d\TG:i:s\Z");
173+
$output_file = (new DataResource($this))
174+
->setURI("http://download.bio2rdf.org/release/$bVersion/$prefix/")
175+
->setTitle("Bio2RDF v$bVersion RDF version of $prefix (generated at $date)")
176+
->setSource($source_file->getURI())
177+
->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/biomodels/biomodels.php")
178+
->setCreateDate($date)
179+
->setHomepage("http://download.bio2rdf.org/release/$bVersion/$prefix/$prefix.html")
180+
->setPublisher("http://bio2rdf.org")
181+
->setRights("use-share-modify")
182+
->setRights("by-attribution")
183+
->setRights("restricted-by-source-license")
184+
->setLicense("http://creativecommons.org/licenses/by/3.0/")
185+
->setDataset(parent::getDatasetURI());
186+
187+
if($gz) $output_file->setFormat("application/gzip");
188+
if(strstr(parent::getParameterValue('output_format'),"nt")) $output_file->setFormat("application/n-triples");
189+
else $output_file->setFormat("application/n-quads");
174190

191+
$dataset_description .= $output_file->toRDF();
192+
193+
//write dataset description to file
194+
parent::setGraphURI($graph_uri);
195+
parent::setWriteFile($odir.parent::getBio2RDFReleaseFile());
196+
parent::getWriteFile()->write($dataset_description);
197+
parent::getWriteFile()->close();
198+
echo "done!".PHP_EOL;
199+
175200
return true;
176201
}
177-
178202
}
179-
$start = microtime(true);
180-
181-
set_error_handler('error_handler');
182-
$parser = new BiomodelsParser($argv);
183-
$parser->Run();
184203

185-
$end = microtime(true);
186-
$time_taken = $end - $start;
187-
print "Started: ".date("l jS F \@ g:i:s a", $start)."\n";
188-
print "Finished: ".date("l jS F \@ g:i:s a", $end)."\n";
189-
print "Took: ".$time_taken." seconds\n"
190204
?>

0 commit comments

Comments
 (0)