handle file names in provenance and change to structure

Author: wow-such-code

src/main/groovy/life/qbic/registration/MainETL.groovy

@@ -30,16 +30,4 @@ class MainETL extends AbstractJavaDataSetRegistrationDropboxV2 {
         }
     }
 
-    private static Optional<String> getExtractedRootDir(String tmpFolderPath) {
-        File tmpDir = new File(tmpFolderPath)
-        String[] childList = tmpDir.list();
-        if(childList.size() == 1) {
-            String childPath = tmpDir.getAbsolutePath() + "/" + childList[0];
-            if (new File(childPath).isDirectory()) {
-                return Optional.of(childPath)
-            }
-        }
-        return Optional.empty()
-    }
-
 }

src/main/groovy/life/qbic/registration/Provenance.groovy

@@ -2,17 +2,34 @@ package life.qbic.registration
 
 import groovy.json.JsonSlurper
 
+/**
+ * Example of a provenance file:
+ * {
+ "origin": "/Users/myuser/registration",
+ "user": "/Users/myuser",
+ "measurementId": "NGSQTEST001AE-1234512312",
+ "datasetFiles" : [ "file1_1.fastq.gz", "file1_2.fastq.gz" ],
+ "taskId": "74c5d26f-b756-42c3-b6f4-2b4825670a2d",
+ "history": [
+ "/opt/scanner-app/scanner-processing-dir/74c5d26f-b756-42c3-b6f4-2b4825670a2d"
+ ]
+ }
+ */
+
 class Provenance {
 
     final String origin
     final String user
     final String measurementID
+    final List<String> datasetFiles
     final List<String> history
 
-    Provenance(String origin, String user, String measurementID, List<String> history) {
+    Provenance(String origin, String user, String measurementID, List<String> datasetFiles,
+               List<String> history) {
         this.origin = origin
         this.user = user
         this.measurementID = measurementID
+        this.datasetFiles = datasetFiles
         this.history = history
     }
 
@@ -21,7 +38,9 @@ class Provenance {
         def object = jsonSlurper.parseText(jsonString)
         assert object instanceof Map
         assert object.history instanceof List
+        assert object.datasetFiles instanceof List
         return new Provenance(object.origin as String, object.user as String,
-                object.measurementId as String, object.history as List<String>)
+                object.measurementId as String, object.datasetFiles as List<String>,
+                object.history as List<String>)
     }
 }

src/main/groovy/life/qbic/registration/handler/DatasetLocatorImpl.groovy

@@ -38,31 +38,26 @@ class DatasetLocatorImpl implements DatasetLocator{
      */
     @Override
     String getPathToDataset() {
-        return findNestedDataset().orElseThrow()
+        return findOrCreateDataPath().orElseThrow()
     }
 
-    private Optional<String> findNestedDataset() {
+    private Optional<String> findOrCreateDataPath() {
         File rootDir = new File(incomingPath)
         if (!rootDir.isDirectory())
             return Optional.empty()
-        /*
-        Now we iterate through the list of child elements in the folder, and search for the directory name appearances.
-        If it is a child directory, that means it has been processed with the dropboxhandler and
-        we can return the nested dataset path.
-        */
-        for (String child : rootDir.list()) {
-            if (child.startsWith(provenance.measurementID)) {
-                String innerFolderPath = rootDir.getAbsolutePath() + "/" + child
-                File innerFolder = new File(innerFolderPath)
-                if(innerFolder.list().size() == 1) {
-                    String singleFilePath = innerFolderPath+"/"+innerFolder.list()[0]
-                    return Optional.of(singleFilePath)
-                } else {
-                    return Optional.of(innerFolderPath)
-                }
-            }
+        List<String> fileNames = provenance.datasetFiles;
+        if(fileNames.size() == 0)
+            return Optional.empty()
+        if(fileNames.size() == 1)
+            return Optional.of(rootDir.getAbsolutePath() + "/" + fileNames[0])
+        // more than one file
+        var newDataFolder = new File(rootDir.getAbsolutePath(), "data");
+        newDataFolder.mkdir();
+        for(String fileName : fileNames) {
+            String targetPath = newDataFolder.getAbsolutePath()+"/"+fileName
+            new File(rootDir.getAbsolutePath()+"/"+fileName).renameTo(targetPath)
         }
-        return Optional.empty()
+        return Optional.of(newDataFolder.getAbsolutePath())
     }
 
     private static boolean isAbsolutePath(String path) {