Changes to make the BII handle external URIs better. Requires further…

… testing by @proccaserra #49

model/src/main/java/uk/ac/ebi/bioinvindex/search/hibernatesearch/bridge/AssayBridge.java

@@ -47,10 +47,8 @@
 import org.apache.lucene.document.Field;
 import org.hibernate.search.bridge.FieldBridge;
 import org.hibernate.search.bridge.LuceneOptions;
-import uk.ac.ebi.bioinvindex.model.Annotation;
 import uk.ac.ebi.bioinvindex.model.AssayResult;
 import uk.ac.ebi.bioinvindex.model.processing.Assay;
-import uk.ac.ebi.bioinvindex.model.term.AnnotationTypes;
 import uk.ac.ebi.bioinvindex.model.xref.Xref;
 import uk.ac.ebi.bioinvindex.search.hibernatesearch.StudyBrowseField;
 import uk.ac.ebi.bioinvindex.utils.datasourceload.DataLocationManager;
@@ -81,31 +79,7 @@ public void set(String s, Object o, Document document, LuceneOptions luceneOptio
         dataLocationManager.setEntityManager(entityManager);
 
         for (Assay assay : assays) {
-
-            if (assay.getTechnologyName().equals("mass spectrometry")) {
-                Collection<AssayResult> assayResults = ProcessingUtils.findAssayResultsFromAssay(assay);
-
-                String fileLink = dataLocationManager.getDataLocationLink(assay.getMeasurement().getName(), assay.getTechnologyName(), assay.getStudy().getObfuscationCode(),
-                        AnnotationTypes.GENERIC_DATA_FILE_LINK);
-
-                System.out.println("File link: " + fileLink);
-
-                String pathLink = dataLocationManager.getDataLocationLink(assay.getMeasurement().getName(), assay.getTechnologyName(), assay.getStudy().getObfuscationCode(),
-                        AnnotationTypes.GENERIC_DATA_FILE_PATH);
-
-                System.out.println("Path link: " + pathLink);
-
-                for (AssayResult result : assayResults) {
-                    for (Annotation annotation : result.getData().getAnnotation("metaboliteFile")) {
-                        System.out.printf("Type: %s -> Value: %s\n", annotation.getType().getValue(), annotation.getText());
-
-                        MetaboLightsIndexer.indexMetaboliteFile(pathLink.replace("${study-acc}",
-                                assay.getStudy().getObfuscationCode()) + annotation.getText(),
-                                document,
-                                luceneOptions);
-                    }
-                }
-            }
+            Collection<AssayResult> assayResults = ProcessingUtils.findAssayResultsFromAssay(assay);
 
             String type = buildType(assay);
 
@@ -114,19 +88,11 @@ public void set(String s, Object o, Document document, LuceneOptions luceneOptio
                 assayTypeToInfo.put(type, info);
             }
 
-            for (Xref xref : assay.getXrefs()) {
-                System.out.println("Adding XREF to AssayTypeInfo: " + xref.getSource().getAcc() + "(" + xref.getAcc() + ") for " + type);
-
-                StringBuilder sb = new StringBuilder();
-                sb.append("xref(").append(xref.getAcc()).append("->");
-                sb.append(xref.getSource().getAcc()).append(")");
-
-                assayTypeToInfo.get(type).addAccession(sb.toString());
-            }
+            createAssayExternalLinks(assayTypeToInfo, assayResults, type);
+            createXrefs(assayTypeToInfo, assay, type);
 
             assayTypeToInfo.get(type).increaseCount();
         }
-        // each data link should be stored perhaps, or at least whatever is required to make it display in the Study page.
 
         for (String type : assayTypeToInfo.keySet()) {
             StringBuilder fullInfo = new StringBuilder();
@@ -160,6 +126,35 @@ public void set(String s, Object o, Document document, LuceneOptions luceneOptio
         entityManager.close();
     }
 
+    private void createXrefs(Map<String, AssayTypeInfo> assayTypeToInfo, Assay assay, String type) {
+        for (Xref xref : assay.getXrefs()) {
+            StringBuilder sb = new StringBuilder();
+            sb.append("xref(").append(xref.getAcc()).append("->");
+            sb.append(xref.getSource().getAcc()).append(")");
+            assayTypeToInfo.get(type).addAccession(sb.toString());
+        }
+    }
+
+    private void createAssayExternalLinks(Map<String, AssayTypeInfo> assayTypeToInfo, Collection<AssayResult> assayResults, String type) {
+        Set<String> addedLinks = new HashSet<String>();
+        for (AssayResult result : assayResults) {
+            // we're only looking at links...should accommodate webdav etc. too
+            //todo remove this after testing
+            if (!result.getData().getName().matches("(http|ftp|https).*") && result.getData().getName().contains("/")) {
+                // we only store the folder since that will take us to multiple file locations. Otherwise we'd have too
+                // many individual links pointing to the same place.
+                String folder = result.getData().getName().substring(0, result.getData().getName().lastIndexOf("/"));
+                if (!addedLinks.contains(folder)) {
+                    StringBuilder sb = new StringBuilder();
+                    sb.append("link(").append(folder).append("->");
+                    sb.append(result.getData().getType().getName()).append(")");
+                    addedLinks.add(folder);
+                    assayTypeToInfo.get(type).addAccession(sb.toString());
+                }
+            }
+        }
+    }
+
     private String buildType(Assay assay) {
         return assay.getMeasurement().getName() + "|" + assay.getTechnologyName();
     }

model/src/main/java/uk/ac/ebi/bioinvindex/utils/datasourceload/Repository.java

@@ -11,7 +11,8 @@
 public enum Repository {
 
     PRIDE("^[0-9]+","pride"), ARRAYEXPRESS("^(e-).*", "arrayexpress"),  GEO("^[(gse)|(ges)]*[0-9]+", "geo"),
-    ENA("^[(sra)|(ena)]*[0-9]+","embl-bank", "ena", "ebi/ena"), GENERIC("generic", "generic", "none", "generic proteomic location", "generic microarray location");
+    ENA("^[(sra)|(ena)]*[0-9]+","embl-bank", "ena", "ebi/ena"),
+    GENERIC("generic", "generic", "none", "generic proteomic location", "generic microarray location");
 
 
     private String startString;

services/src/main/java/uk/ac/ebi/bioinvindex/services/SourceURLResolverImpl.java

@@ -65,6 +65,8 @@
 @AutoCreate
 public class SourceURLResolverImpl implements SourceURLResolver {
 
+    public static final String EXTERNAL_LINK_PATTERN = "(http|ftp|https).*";
+
     private static Cache<String, String> cache = new BIICache<String, String>();
 
     private static final Log log = LogFactory.getLog(SourceURLResolverImpl.class);
@@ -74,16 +76,21 @@ public class SourceURLResolverImpl implements SourceURLResolver {
 
     public String getRawDataURL(String measurement, String technology, String accession) {
         try {
+            if (checkIsExternalLink(accession)) return accession;
             return getDataURL(measurement, technology, accession, AnnotationTypes.RAW_DATA_FILE_LINK);
         } catch (Exception e) {
             log.error("Unable to resolve Raw data URL");
             return "";
         }
     }
 
+    private boolean checkIsExternalLink(String accession) {
+        return accession.matches(EXTERNAL_LINK_PATTERN);
+    }
+
     public String getProcessedDataURL(String measurement, String technology, String accession) {
         try {
-
+            if (checkIsExternalLink(accession)) return accession;
             return getDataURL(measurement, technology, accession, AnnotationTypes.PROCESSED_DATA_FILE_LINK);
         } catch (Exception e) {
             log.error("Unable to resolve Processed data URL");

services/src/main/java/uk/ac/ebi/bioinvindex/services/StudyBeanImpl.java

@@ -94,23 +94,11 @@ public class StudyBeanImpl implements StudyBean {
     @In(required = false)
     private StudyIndexLocatorImpl studyIndexLocator;
 
-
-    private String organism;
-
-    private String design;
-
-    private Map<String, List<String>> factorsToValues;
-
-    private Map<String, List<String>> characteristicsToValues;
-
-    private List<Contact> contacts;
-
-    private List<AssayGroupInfo> assayInfos;
-
-    private String studyId;
-
+    private String studyId, organism, design;
     private Collection<Investigation> investigations;
-
+    private List<AssayGroupInfo> assayInfos;
+    private List<Contact> contacts;
+    private Map<String, List<String>> characteristicsToValues, factorsToValues;
     private List<String> relatedStudies;
 
     public StudyBeanImpl() {

services/src/main/java/uk/ac/ebi/bioinvindex/services/browse/BrowseStudyBeanImpl.java

@@ -354,12 +354,12 @@ private AssayInfoBean createAssayInfoBean(AssayInfoBean assayInfoBean, String as
     private DataLink createDBLink(String dbLinkRepresentation) {
         DataLink link = new DataLink();
 
-        dbLinkRepresentation = dbLinkRepresentation.replace("xref(", "").replace(")", "");
+        dbLinkRepresentation = dbLinkRepresentation.replace("xref(", "").replace("link(", "").replace(")", "");
 
         if (dbLinkRepresentation.contains("->")) {
             String[] dbLinkParts = dbLinkRepresentation.split("->");
             link.setAcc(dbLinkParts[0]);
-            link.setSourceName(dbLinkParts[1]);
+            link.setSourceName(dbLinkParts[1].toUpperCase());
         }
 
         return link;