Users can create and analyze datasets. (#12)

* Add dropdown component * adding ui components * adding datasetName to API * adding datasetName to API * adding datasetName to API * adding datasetName to API * adding datasets to UI * Add dropdown component * Add dropdown component * Improve Dataset selection --------- Co-authored-by: artur <[email protected]> Co-authored-by: artur <[email protected]>
datascience · Jul 15, 2024 · e196165 · e196165
1 parent ff72ed3
commit e196165
Show file tree

Hide file tree

Showing 38 changed files with 550 additions and 291 deletions.
diff --git a/core/src/main/java/rocks/artur/api/AnalyzePersistFile.java b/core/src/main/java/rocks/artur/api/AnalyzePersistFile.java
@@ -11,8 +11,8 @@
  */
 public interface AnalyzePersistFile {
 
-    Long uploadCharacterisationResults(ByteFile file);
+    Long uploadCharacterisationResults(ByteFile file, String datasetName);
 
-    Long uploadCharacterisationResults(List<ByteFile> files);
+    Long uploadCharacterisationResults(List<ByteFile> files, String datasetName);
 
 }
diff --git a/core/src/main/java/rocks/artur/api/GetCollectionStatistics.java b/core/src/main/java/rocks/artur/api/GetCollectionStatistics.java
@@ -5,6 +5,6 @@
 import java.util.Map;
 
 public interface GetCollectionStatistics {
-    Map<String, Double> getStatistics(FilterCriteria filterCriteria);
+    Map<String, Double> getStatistics(FilterCriteria filterCriteria, String datasetName);
 
 }
diff --git a/core/src/main/java/rocks/artur/api/GetDatasetInfo.java b/core/src/main/java/rocks/artur/api/GetDatasetInfo.java
@@ -0,0 +1,7 @@
+package rocks.artur.api;
+
+import java.util.List;
+
+public interface GetDatasetInfo {
+    List<String> listDatasets();
+}
diff --git a/core/src/main/java/rocks/artur/api/GetObjects.java b/core/src/main/java/rocks/artur/api/GetObjects.java
@@ -10,8 +10,8 @@
  * This interface enables getting characterisation results.
  */
 public interface GetObjects {
-    List<PropertiesPerObjectStatistic> getObjects(FilterCriteria filterCriteria);
-    Iterable<CharacterisationResult> getObject(String filePath);
+    List<PropertiesPerObjectStatistic> getObjects(FilterCriteria filterCriteria, String datasetName);
+    Iterable<CharacterisationResult> getObject(String filePath, String datasetName);
 
-    List<CharacterisationResult> getConflictsFromObject(String filePath);
+    List<CharacterisationResult> getConflictsFromObject(String filePath, String datasetName);
 }
diff --git a/core/src/main/java/rocks/artur/api/GetProperties.java b/core/src/main/java/rocks/artur/api/GetProperties.java
@@ -9,6 +9,6 @@
  * This interface enables getting a property distribution.
  */
 public interface GetProperties {
-    List<PropertyStatistic> getProperties();
-    List<PropertyStatistic> getProperties(FilterCriteria filter);
+    List<PropertyStatistic> getProperties(String datasetName);
+    List<PropertyStatistic> getProperties(FilterCriteria filter, String datasetName);
 }
diff --git a/core/src/main/java/rocks/artur/api/GetPropertyValueDistribution.java b/core/src/main/java/rocks/artur/api/GetPropertyValueDistribution.java
@@ -11,5 +11,5 @@
  * This interface enables getting a property value distribution given a property name.
  */
 public interface GetPropertyValueDistribution {
-    List<PropertyValueStatistic> getPropertyValueDistribution(Property propertyName, FilterCriteria<CharacterisationResult> filterCriteria);
+    List<PropertyValueStatistic> getPropertyValueDistribution(Property propertyName, FilterCriteria<CharacterisationResult> filterCriteria, String datasetName);
 }
diff --git a/core/src/main/java/rocks/artur/api/GetPropertyValueDistributionWithFilter.java b/core/src/main/java/rocks/artur/api/GetPropertyValueDistributionWithFilter.java
@@ -8,5 +8,5 @@
  * This interface enables getting a property value distribution given a filter.
  */
 public interface GetPropertyValueDistributionWithFilter {
-    List<PropertyValueStatistic> getPropertyValueDistributionWithFilter(String propertyName, String filter);
+    List<PropertyValueStatistic> getPropertyValueDistributionWithFilter(String propertyName, String filter, String datasetName);
 }
diff --git a/core/src/main/java/rocks/artur/api/GetSamples.java b/core/src/main/java/rocks/artur/api/GetSamples.java
@@ -15,7 +15,7 @@ public interface GetSamples {
 
     void setProperties(List<Property> properties);
 
-    Iterable<String> getObjects(FilterCriteria filterCriteria);
+    Iterable<String> getObjects(FilterCriteria filterCriteria, String datasetName);
 
-    List<String[]> getSamplingInfo(FilterCriteria filterCriteria);
+    List<String[]> getSamplingInfo(FilterCriteria filterCriteria, String datasetName);
 }
diff --git a/core/src/main/java/rocks/artur/api/GetSources.java b/core/src/main/java/rocks/artur/api/GetSources.java
@@ -6,5 +6,5 @@
  * This interface enables getting a property distribution.
  */
 public interface GetSources {
-    List<String> getSources();
+    List<String> getSources(String datasetName);
 }
diff --git a/core/src/main/java/rocks/artur/api/ResolveConflicts.java b/core/src/main/java/rocks/artur/api/ResolveConflicts.java
@@ -1,5 +1,5 @@
 package rocks.artur.api;
 
 public interface ResolveConflicts {
-    void run();
+    void run(String datasetName);
 }
diff --git a/core/src/main/java/rocks/artur/api_impl/AnalyzePersistFileImpl.java b/core/src/main/java/rocks/artur/api_impl/AnalyzePersistFileImpl.java
@@ -21,20 +21,20 @@ public AnalyzePersistFileImpl(CharacterisationResultProducer characterisationRes
     }
 
     @Override
-    public Long uploadCharacterisationResults(ByteFile file) {
+    public Long uploadCharacterisationResults(ByteFile file, String datasetName) {
         List<CharacterisationResult> characterisationResults = characterisationResultProducer.processFile(file);
-        characterisationResultGateway.addCharacterisationResults(characterisationResults);
+        characterisationResultGateway.addCharacterisationResults(characterisationResults, datasetName);
         return Long.valueOf(characterisationResults.size());
     }
 
     @Override
-    public Long uploadCharacterisationResults(List<ByteFile> files) {
+    public Long uploadCharacterisationResults(List<ByteFile> files, String datasetName) {
         List<CharacterisationResult> characterisationResults = new ArrayList<>();
         files.stream().forEach(file -> {
             List<CharacterisationResult> tmp = characterisationResultProducer.processFile(file);
             characterisationResults.addAll(tmp);
         });
-        characterisationResultGateway.addCharacterisationResults(characterisationResults);
+        characterisationResultGateway.addCharacterisationResults(characterisationResults, datasetName);
         return Long.valueOf(characterisationResults.size());
     }
 }
diff --git a/core/src/main/java/rocks/artur/api_impl/CRH_ResolveConflictsImpl.java b/core/src/main/java/rocks/artur/api_impl/CRH_ResolveConflictsImpl.java
@@ -22,41 +22,41 @@ public CRH_ResolveConflictsImpl(CharacterisationResultGateway characterisationRe
     }
 
 
-    public void run() {
-        init();
+    public void run(String datasetName) {
+        init(datasetName);
         System.out.println(sourceWeights);
         //System.out.println("sum of weights: " +  sourceWeights.values().stream().reduce(0d, Double::sum));
-        updateTruth();
+        updateTruth(datasetName);
         System.out.println("sum of weights: " +  sourceWeights.values().stream().reduce(0d, Double::sum));
         //System.out.println(truth);
         for (int i = 0; i < 3; i++) {
-            updateWeights();
+            updateWeights(datasetName);
             System.out.println(sourceWeights);
             System.out.println("sum of weights: " +  sourceWeights.values().stream().reduce(0d, Double::sum));
-            updateTruth();
+            updateTruth(datasetName);
             //System.out.println(truth);
         }
 
-        resolveConflicts();
+        resolveConflicts(datasetName);
     }
 
-    private void resolveConflicts() {
+    private void resolveConflicts(String datasetName) {
         truth.entrySet().stream().forEach( entry -> {
             Entry key = entry.getKey();
             String value = entry.getValue();
 
-            List<CharacterisationResult> characterisationResultsByEntry = characterisationResultGateway.getCharacterisationResultsByEntry(key);
+            List<CharacterisationResult> characterisationResultsByEntry = characterisationResultGateway.getCharacterisationResultsByEntry(key, datasetName);
             for (CharacterisationResult characterisationResult : characterisationResultsByEntry) {
                 if (!characterisationResult.getValue().equals(value)) {
-                    characterisationResultGateway.delete(characterisationResult);
+                    characterisationResultGateway.delete(characterisationResult, datasetName);
                 }
             }
 
 
         });
     }
 
-    private void updateWeights() {
+    private void updateWeights(String datasetName) {
         Map<String, Double> score = sources.stream().collect(Collectors.toMap(
                 Function.identity(),
                 s -> 0.0));
@@ -66,10 +66,10 @@ private void updateWeights() {
                 s -> 0.0));
 
 
-        List<Entry> entries = characterisationResultGateway.getEntries();
+        List<Entry> entries = characterisationResultGateway.getEntries(datasetName);
 
         for (Entry entry : entries) {
-            List<CharacterisationResult> characterisationResults = characterisationResultGateway.getCharacterisationResultsByEntry(entry);
+            List<CharacterisationResult> characterisationResults = characterisationResultGateway.getCharacterisationResultsByEntry(entry, datasetName);
 
             for (CharacterisationResult characterisationResult : characterisationResults) {
 
@@ -112,10 +112,10 @@ private void updateWeights() {
         }
     }
 
-    private void updateTruth() {
-        List<Entry> entries = characterisationResultGateway.getEntries();
+    private void updateTruth(String datasetName) {
+        List<Entry> entries = characterisationResultGateway.getEntries(datasetName);
         for (Entry entry : entries) {
-            List<CharacterisationResult> characterisationResults = characterisationResultGateway.getCharacterisationResultsByEntry(entry);
+            List<CharacterisationResult> characterisationResults = characterisationResultGateway.getCharacterisationResultsByEntry(entry, datasetName);
 
             if (characterisationResults.size() > 0) {
                 CharacterisationResult firstResult = characterisationResults.get(0);
@@ -140,9 +140,9 @@ private void updateTruth() {
     Map<String, Double> sourceWeights;
     Map<Entry, String> truth;
 
-    void init() {
+    void init(String datasetName) {
 
-        sources = characterisationResultGateway.getSources();
+        sources = characterisationResultGateway.getSources(datasetName);
         sourceWeights = sources.stream().collect(Collectors.toMap(
                 Function.identity(),
                 s -> 1.0 / sources.size()));

diff --git a/core/src/main/java/rocks/artur/api_impl/GetCollectionStatisticsImpl.java b/core/src/main/java/rocks/artur/api_impl/GetCollectionStatisticsImpl.java
@@ -15,8 +15,8 @@ public GetCollectionStatisticsImpl(CharacterisationResultGateway characterisatio
     }
 
     @Override
-    public Map<String, Double> getStatistics(FilterCriteria filterCriteria) {
-        Map<String, Double> sizeStatistics = characterisationResultGateway.getCollectionStatistics(filterCriteria);
+    public Map<String, Double> getStatistics(FilterCriteria filterCriteria, String datasetName) {
+        Map<String, Double> sizeStatistics = characterisationResultGateway.getCollectionStatistics(filterCriteria, datasetName);
         return sizeStatistics;
     }
 

diff --git a/core/src/main/java/rocks/artur/api_impl/GetDatasetInfoImpl.java b/core/src/main/java/rocks/artur/api_impl/GetDatasetInfoImpl.java
@@ -0,0 +1,19 @@
+package rocks.artur.api_impl;
+
+import rocks.artur.api.GetDatasetInfo;
+import rocks.artur.domain.CharacterisationResultGateway;
+
+import java.util.List;
+
+public class GetDatasetInfoImpl implements GetDatasetInfo {
+    private CharacterisationResultGateway characterisationResultGateway;
+
+    public GetDatasetInfoImpl(CharacterisationResultGateway characterisationResultGateway) {
+        this.characterisationResultGateway = characterisationResultGateway;
+    }
+
+    @Override
+    public List<String> listDatasets() {
+        return this.characterisationResultGateway.listDatasets();
+    }
+}
diff --git a/core/src/main/java/rocks/artur/api_impl/GetObjectsImpl.java b/core/src/main/java/rocks/artur/api_impl/GetObjectsImpl.java
@@ -16,20 +16,20 @@ public GetObjectsImpl(CharacterisationResultGateway characterisationResultGatewa
     }
 
     @Override
-    public List<PropertiesPerObjectStatistic> getObjects(FilterCriteria filterCriteria) {
-        List<PropertiesPerObjectStatistic> objects = characterisationResultGateway.getObjects(filterCriteria);
+    public List<PropertiesPerObjectStatistic> getObjects(FilterCriteria filterCriteria, String datasetName) {
+        List<PropertiesPerObjectStatistic> objects = characterisationResultGateway.getObjects(filterCriteria, datasetName);
         return objects;
     }
 
     @Override
-    public Iterable<CharacterisationResult> getObject(String filePath) {
-        Iterable<CharacterisationResult> characterisationResultsByFilepath = characterisationResultGateway.getCharacterisationResultsByFilepath(filePath);
+    public Iterable<CharacterisationResult> getObject(String filePath, String datasetName) {
+        Iterable<CharacterisationResult> characterisationResultsByFilepath = characterisationResultGateway.getCharacterisationResultsByFilepath(filePath, datasetName);
         return characterisationResultsByFilepath;
     }
 
     @Override
-    public List<CharacterisationResult> getConflictsFromObject(String filePath) {
-        List<CharacterisationResult> characterisationResultsByFilepath = characterisationResultGateway.getConflictsByFilepath(filePath);
+    public List<CharacterisationResult> getConflictsFromObject(String filePath, String datasetName) {
+        List<CharacterisationResult> characterisationResultsByFilepath = characterisationResultGateway.getConflictsByFilepath(filePath, datasetName);
         return characterisationResultsByFilepath;
     }
 

diff --git a/core/src/main/java/rocks/artur/api_impl/GetPropertiesImpl.java b/core/src/main/java/rocks/artur/api_impl/GetPropertiesImpl.java
@@ -15,14 +15,14 @@ public GetPropertiesImpl(CharacterisationResultGateway characterisationResultGat
     }
 
     @Override
-    public List<PropertyStatistic> getProperties() {
-        List<PropertyStatistic> propertyDistribution = characterisationResultGateway.getPropertyDistribution(null);
+    public List<PropertyStatistic> getProperties(String datasetName) {
+        List<PropertyStatistic> propertyDistribution = characterisationResultGateway.getPropertyDistribution(null, datasetName);
         return propertyDistribution;
     }
 
     @Override
-    public List<PropertyStatistic> getProperties(FilterCriteria filter) {
-        List<PropertyStatistic> propertyDistribution = characterisationResultGateway.getPropertyDistribution(filter);
+    public List<PropertyStatistic> getProperties(FilterCriteria filter, String datasetName) {
+        List<PropertyStatistic> propertyDistribution = characterisationResultGateway.getPropertyDistribution(filter, datasetName);
         return propertyDistribution;
     }
 }
diff --git a/core/src/main/java/rocks/artur/api_impl/GetPropertyValueDistributionImpl.java b/core/src/main/java/rocks/artur/api_impl/GetPropertyValueDistributionImpl.java
@@ -18,8 +18,8 @@ public GetPropertyValueDistributionImpl(CharacterisationResultGateway characteri
 
 
     @Override
-    public List<PropertyValueStatistic> getPropertyValueDistribution(Property property, FilterCriteria<CharacterisationResult> filterCriteria) {
-        List<PropertyValueStatistic> valueDistributionByProperty = characterisationResultGateway.getPropertyValueDistribution(property, filterCriteria);
+    public List<PropertyValueStatistic> getPropertyValueDistribution(Property property, FilterCriteria<CharacterisationResult> filterCriteria, String datasetName) {
+        List<PropertyValueStatistic> valueDistributionByProperty = characterisationResultGateway.getPropertyValueDistribution(property, filterCriteria, datasetName);
         return valueDistributionByProperty;
     }
 }
diff --git a/core/src/main/java/rocks/artur/api_impl/GetSamplesImpl.java b/core/src/main/java/rocks/artur/api_impl/GetSamplesImpl.java
@@ -34,8 +34,8 @@ public void setProperties(List<Property> properties) {
     }
 
     @Override
-    public List<String> getObjects(FilterCriteria filterCriteria) {
-        List<String[]> samplingResults = characterisationResultGateway.getSamples(filterCriteria, algorithm, properties);
+    public List<String> getObjects(FilterCriteria filterCriteria, String datasetName) {
+        List<String[]> samplingResults = characterisationResultGateway.getSamples(filterCriteria, algorithm, properties, datasetName);
         List<String> results = new ArrayList<>();
         switch (algorithm) {
 
@@ -50,8 +50,8 @@ public List<String> getObjects(FilterCriteria filterCriteria) {
     }
 
     @Override
-    public List<String[]> getSamplingInfo(FilterCriteria filterCriteria) {
-        List<String[]> samplingResults = characterisationResultGateway.getSamples(filterCriteria, algorithm, properties);
+    public List<String[]> getSamplingInfo(FilterCriteria filterCriteria, String datasetName) {
+        List<String[]> samplingResults = characterisationResultGateway.getSamples(filterCriteria, algorithm, properties, datasetName);
 
         return samplingResults;
     }

diff --git a/core/src/main/java/rocks/artur/api_impl/GetSourcesImpl.java b/core/src/main/java/rocks/artur/api_impl/GetSourcesImpl.java
@@ -13,8 +13,8 @@ public GetSourcesImpl(CharacterisationResultGateway characterisationResultGatewa
     }
 
     @Override
-    public List<String> getSources() {
-        List<String> sources = characterisationResultGateway.getSources();
+    public List<String> getSources(String datasetName) {
+        List<String> sources = characterisationResultGateway.getSources(datasetName);
         return sources;
     }
 }
diff --git a/core/src/main/java/rocks/artur/api_impl/Native_ResolveConflictsImpl.java b/core/src/main/java/rocks/artur/api_impl/Native_ResolveConflictsImpl.java
@@ -10,7 +10,7 @@ public Native_ResolveConflictsImpl(CharacterisationResultGateway characterisatio
         this.characterisationResultGateway = characterisationResultGateway;
     }
     @Override
-    public void run() {
-        characterisationResultGateway.resolveConflictsNative();
+    public void run(String datasetName) {
+        characterisationResultGateway.resolveConflictsNative(datasetName);
     }
 }
diff --git a/core/src/main/java/rocks/artur/api_impl/filter/FilterOperation.java b/core/src/main/java/rocks/artur/api_impl/filter/FilterOperation.java
@@ -1,6 +1,5 @@
 package rocks.artur.api_impl.filter;
 
-import rocks.artur.domain.ValueType;
 
 public enum FilterOperation {
     LESS("<"), LESS_OR_EQUAL ("<="),