@@ -166,17 +166,20 @@ function parse_dir(){
166
166
$ indir = parent ::getParameterValue ('indir ' );
167
167
if ($ handle = opendir ($ indir )) {
168
168
echo "Processing directory $ indir \n" ;
169
+
170
+ $ outfile = "clinicaltrials. " .parent ::getParameterValue ('output_format ' );
171
+ $ gz = (strstr (parent ::getParameterValue ('output_format ' ),".gz " ) === FALSE )?false :true ;
172
+ parent ::setWriteFile (parent ::getParameterValue ("outdir " ).$ outfile ,$ gz );
173
+
169
174
while (($ file = readdir ($ handle )) !== false ){
170
175
if (in_array ($ file ,$ ignore ) || is_dir ($ file ) ) continue ;
171
176
$ trial_id = basename ($ file ,'.xml ' );
172
177
if (parent ::getParameterValue ('id_list ' ) == '' || in_array ($ trial_id , $ ids )) {
173
178
echo "Processing $ file " .PHP_EOL ;
174
179
$ this ->process_file ($ file );
175
-
176
- $ outfile = basename ($ file ,".xml " ).'. ' .parent ::getParameterValue ('output_format ' );
177
-
180
+
178
181
// make the dataset description
179
- $ ouri = parent ::getGraphURI (parent :: getDatasetURI () );
182
+ $ ouri = parent ::getGraphURI ();
180
183
parent ::setGraphURI (parent ::getDatasetURI ());
181
184
182
185
$ rfile = "http://clinicaltrials.gov/ct2/show/ " .$ trial_id ."?resultsxml=true " ;
@@ -194,31 +197,34 @@ function parse_dir(){
194
197
->setLicense ("http://clinicaltrials.gov/ct2/about-site/terms-conditions " )
195
198
->setDataset ("http://identifiers.org/clinicaltrials/ " );
196
199
197
- $ output_file = (new DataResource ($ this ))
198
- ->setURI ("http://download.bio2df.org/release/ $ bVersion/ $ prefix/ $ outfile " )
199
- ->setTitle ("Bio2RDF v $ bVersion RDF version of $ prefix v $ source_version " )
200
- ->setSource ($ source_file ->getURI ())
201
- ->setCreator ("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/clinicaltrials/clinicaltrials.php " )
202
- ->setCreateDate ($ date )
203
- ->setHomepage ("http://download.bio2rdf.org/release/ $ bVersion/ $ prefix/ $ prefix.html " )
204
- ->setPublisher ("http://bio2rdf.org " )
205
- ->setRights ("use-share-modify " )
206
- ->setRights ("by-attribution " )
207
- ->setRights ("restricted-by-source-license " )
208
- ->setLicense ("http://creativecommons.org/licenses/by/3.0/ " )
209
- ->setDataset (parent ::getDatasetURI ());
210
-
211
- $ gz = (strstr (parent ::getParameterValue ('output_format ' ),".gz " ) === FALSE )?false :true ;
212
- if ($ gz ) $ output_file ->setFormat ("application/gzip " );
213
- if (strstr (parent ::getParameterValue ('output_format ' ),"nt " )) $ output_file ->setFormat ("application/n-triples " );
214
- else $ output_file ->setFormat ("application/n-quads " );
215
-
216
- fwrite ($ fp , $ source_file ->toRDF ().$ output_file ->toRDF ());
217
- parent ::setGraphURI (parent ::setDatasetURI ($ ouri ));
200
+ parent ::writeToReleaseFile ($ source_file ->toRDF ());
201
+ parent ::setGraphURI ($ ouri );
218
202
}
219
203
}
220
- echo "Finished \n . " ;
204
+ echo "Finished. " . PHP_EOL ;
221
205
closedir ($ handle );
206
+
207
+ $ output_file = (new DataResource ($ this ))
208
+ ->setURI ("http://download.bio2df.org/release/ $ bVersion/ $ prefix/ $ outfile " )
209
+ ->setTitle ("Bio2RDF v $ bVersion RDF version of $ prefix v $ source_version " )
210
+ ->setSource ($ source_file ->getURI ())
211
+ ->setCreator ("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/clinicaltrials/clinicaltrials.php " )
212
+ ->setCreateDate ($ date )
213
+ ->setHomepage ("http://download.bio2rdf.org/release/ $ bVersion/ $ prefix/ $ prefix.html " )
214
+ ->setPublisher ("http://bio2rdf.org " )
215
+ ->setRights ("use-share-modify " )
216
+ ->setRights ("by-attribution " )
217
+ ->setRights ("restricted-by-source-license " )
218
+ ->setLicense ("http://creativecommons.org/licenses/by/3.0/ " )
219
+ ->setDataset (parent ::getDatasetURI ());
220
+
221
+ $ gz = (strstr (parent ::getParameterValue ('output_format ' ),".gz " ) === FALSE )?false :true ;
222
+ if ($ gz ) $ output_file ->setFormat ("application/gzip " );
223
+ if (strstr (parent ::getParameterValue ('output_format ' ),"nt " )) $ output_file ->setFormat ("application/n-triples " );
224
+ else $ output_file ->setFormat ("application/n-quads " );
225
+
226
+ parent ::writeToReleaseFile ($ output_file ->toRDF ());
227
+ parent ::closeReleaseFile ();
222
228
223
229
// write the dataset description file
224
230
fclose ($ fp );
@@ -230,12 +236,7 @@ function parse_dir(){
230
236
**/
231
237
function process_file ($ infile ) {
232
238
$ indir = parent ::getParameterValue ('indir ' );
233
- $ outfile = parent ::getParameterValue ("outdir " )
234
- .basename ($ infile ,".xml " )
235
- .'. ' .parent ::getParameterValue ('output_format ' );
236
- $ gz = (strstr (parent ::getParameterValue ('output_format ' ),".gz " ) === FALSE )?false :true ;
237
-
238
- $ this ->setWriteFile ($ outfile ,$ gz );
239
+
239
240
$ xml = new CXML ($ indir ,basename ($ infile ));
240
241
$ this ->setCheckPoint ('file ' );
241
242
while ($ xml ->Parse ("clinical_study " ) == TRUE ) {
@@ -924,10 +925,10 @@ function process_file($infile) {
924
925
echo "There was an error parsing the is_fda_regulated element: $ e \n" ;
925
926
}
926
927
928
+ parent ::writeRDFBufferToWriteFile ();
927
929
}
928
930
$ this ->setCheckPoint ('record ' );
929
931
$ this ->setCheckPoint ('dataset ' );
930
- $ this ->getWriteFile ()->close ();
931
932
}
932
933
933
934
function getString ($ xpath ,$ element = null )
0 commit comments