Significantly improve performance with large datasets

Measurements pre-patch: * CPU Time: 1600090 ms * Memory: 117.41 GB * Total time: 26:45 Measurements post-patch: * CPU Time: 827613 ms (-48.3%) * Memory: 166.57 GB (+41.9%) * Total time: 13:50 (-51%) Signed-off-by: Taylor Smock <[email protected]>
JOSM · Mar 22, 2024 · d35dfd9 · d35dfd9
1 parent e5a6f9b
commit d35dfd9
Show file tree

Hide file tree

Showing 6 changed files with 81 additions and 30 deletions.
diff --git a/src/com/vividsolutions/jcs/conflate/polygonmatch/ChainMatcher.java b/src/com/vividsolutions/jcs/conflate/polygonmatch/ChainMatcher.java
@@ -36,6 +36,7 @@
 
 import com.vividsolutions.jump.feature.Feature;
 import com.vividsolutions.jump.feature.FeatureCollection;
+import com.vividsolutions.jump.feature.FeatureDataset;
 
 /**
  * Composes several FeatureMatchers into one. Candidate features are whittled
@@ -68,8 +69,7 @@ public ChainMatcher(FeatureMatcher[] matchers) {
    */
     @Override
   public Matches match(Feature target, FeatureCollection candidates) {
-    Matches survivors = new Matches(
-        candidates.getFeatureSchema(), candidates.getFeatures());
+    Matches survivors = new Matches(new FeatureDataset(candidates.getFeatures(), candidates.getFeatureSchema(), candidates.getEnvelope()));
     for (FeatureMatcher matcher : matchers) {
       survivors = matcher.match(target, survivors);
     }

diff --git a/src/com/vividsolutions/jcs/conflate/polygonmatch/FeatureMatcher.java b/src/com/vividsolutions/jcs/conflate/polygonmatch/FeatureMatcher.java
@@ -58,6 +58,6 @@ public interface FeatureMatcher {
    * @return the matching features, and a score for each. (Implementors should
    * document how they do their scoring).
    */
-  public Matches match(Feature target, FeatureCollection candidates);
+  Matches match(Feature target, FeatureCollection candidates);
 
 }
diff --git a/src/com/vividsolutions/jcs/conflate/polygonmatch/IndependentCandidateMatcher.java b/src/com/vividsolutions/jcs/conflate/polygonmatch/IndependentCandidateMatcher.java
@@ -31,9 +31,12 @@
  */
 package com.vividsolutions.jcs.conflate.polygonmatch;
 
+import java.util.Collections;
+
 import org.locationtech.jts.geom.Geometry;
 import com.vividsolutions.jump.feature.Feature;
 import com.vividsolutions.jump.feature.FeatureCollection;
+import com.vividsolutions.jump.feature.FeatureDataset;
 
 /**
  * Base class of FeatureMatchers that compare the target to each candidate
@@ -46,7 +49,9 @@ public IndependentCandidateMatcher() {
 
     @Override
   public Matches match(Feature target, FeatureCollection candidates) {
-    Matches matches = new Matches(candidates.getFeatureSchema());
+    final FeatureDataset fds = new FeatureDataset(Collections.emptySet(), candidates.getFeatureSchema(),
+            candidates.getEnvelope());
+    final Matches matches = new Matches(fds);
     for (Feature candidate : candidates) {
       double score = match(target.getGeometry(), candidate.getGeometry());
       if (score > 0) { matches.add(candidate, score); }

diff --git a/src/com/vividsolutions/jcs/conflate/polygonmatch/Matches.java b/src/com/vividsolutions/jcs/conflate/polygonmatch/Matches.java
@@ -55,6 +55,7 @@
  */
 public class Matches extends AbstractMap<Feature, Double> implements FeatureCollection, Cloneable {
     private final Set<Map.Entry<Feature, Double>> entrySet = new HashSet<>();
+    private int size;
     /**
      * Creates a Matches object.
      * @param schema metadata applicable to the features that will be stored in
@@ -82,16 +83,26 @@ protected Matches clone() {
     public Matches(FeatureSchema schema, List<Feature> features) {
         // We want to ensure that the dataset won't have a ton of ArrayList#grow calls
         // So we initialize the dataset with all the data
-        this.dataset = new FeatureDataset(features.size(), schema);
+        this.dataset = new FeatureDataset(features.size(), schema, null);
         this.scores = new double[features.size()];
+        addAll(features, 1, true);
+    }
 
-        for (Feature match : features) {
-            add(match, 1);
-        }
+    /**
+     * Creates a Matches object, initialized with the given Dataset.
+     * @param featureDataset The dataset to use for initialization
+     */
+    public Matches(FeatureDataset featureDataset) {
+        // We want to ensure that the dataset won't have a ton of ArrayList#grow calls
+        // So we initialize the dataset with all the data
+        this.scores = new double[featureDataset.size()];
+        this.dataset = new FeatureDataset(featureDataset);
+
+        addAll(featureDataset, 1, false);
     }
 
     private final FeatureDataset dataset;
-    private double[] scores = new double[0];
+    private double[] scores;
 
     /**
      * This method is not supported, because added features need to be associated
@@ -113,6 +124,12 @@ public void addAll(Collection<? extends Feature> features) {
         throw new UnsupportedOperationException("Use #add(feature, score) instead");
     }
 
+    private void addAll(Iterable<? extends Feature> features, double score, boolean addToDataset) {
+        for (Feature feature : features) {
+            add(feature, score, addToDataset);
+        }
+    }
+
     /**
      * This method is not supported, because added features need to be associated
      * with a score. Use #add(Feature, double) instead.
@@ -191,6 +208,10 @@ public void remove(Feature feature) {
      * @param score the confidence of the match, ranging from 0 to 1
      */
     public void add(Feature feature, double score) {
+        add(feature, score, true);
+    }
+
+    private void add(Feature feature, double score, boolean addToDataset) {
         // We want to avoid the string concatenation here, if we don't need it.
         // It is *very* expensive when run with large datasets.
         // This used to be an Assert.isTrue statement
@@ -200,8 +221,8 @@ public void add(Feature feature, double score) {
         if (score == 0) {
             return;
         }
-        scoreAdd(dataset.size(), score);
-        dataset.add(feature);
+        scoreAdd(size++, score);
+        if (addToDataset) dataset.add(feature);
         if (score > topScore) {
             topScore = score;
             topMatch = feature;
@@ -216,7 +237,7 @@ private void scoreAdd(int index, double score) {
     }
 
     private Feature topMatch;
-    private double topScore = 0;
+    private double topScore;
 
     public double getTopScore() {
         return topScore;

diff --git a/src/com/vividsolutions/jcs/conflate/polygonmatch/WindowMatcher.java b/src/com/vividsolutions/jcs/conflate/polygonmatch/WindowMatcher.java
@@ -37,6 +37,7 @@
 import org.locationtech.jts.geom.Envelope;
 import com.vividsolutions.jump.feature.Feature;
 import com.vividsolutions.jump.feature.FeatureCollection;
+import com.vividsolutions.jump.feature.FeatureDataset;
 import com.vividsolutions.jump.geom.EnvelopeUtil;
 
 /**
@@ -79,6 +80,6 @@ public WindowMatcher() {}
   public Matches match(Feature target, FeatureCollection candidates) {
     Envelope window = new Envelope(target.getGeometry().getEnvelopeInternal());
     window = EnvelopeUtil.expand(window, buffer);
-    return new Matches(candidates.getFeatureSchema(), candidates.query(window));
+    return new Matches(new FeatureDataset(candidates.query(window), candidates.getFeatureSchema(), window));
   }
 }
diff --git a/src/com/vividsolutions/jump/feature/FeatureDataset.java b/src/com/vividsolutions/jump/feature/FeatureDataset.java
@@ -55,34 +55,58 @@ public class FeatureDataset implements FeatureCollection {
      * @param featureSchema the types of the attributes of the features in this collection
      */
     public FeatureDataset(Collection<Feature> newFeatures, FeatureSchema featureSchema) {
-        features = new ArrayList<>(newFeatures);
+        this(newFeatures, featureSchema, null);
+    }
+
+    /**
+     * Creates a FeatureDataset, initialized with a group of Features.
+     * @param newFeatures an initial group of features to add to this FeatureDataset
+     * @param featureSchema the types of the attributes of the features in this collection
+     * @param envelope The expected envelope for the features
+     */
+    public FeatureDataset(Collection<Feature> newFeatures, FeatureSchema featureSchema, Envelope envelope) {
+        this.features = new ArrayList<>(newFeatures);
         this.featureSchema = featureSchema;
+        this.envelope = envelope != null ? envelope.copy() : null;
     }
 
     /**
      * Creates a FeatureDataset, with an initial list size for features
      * @param newFeatures The expected size of the features
      * @param featureSchema the types of the attributes of the features in this collection
+     * @param envelope The expected envelope for this dataset
      */
-    public FeatureDataset(int newFeatures, FeatureSchema featureSchema) {
-        features = new ArrayList<>(newFeatures);
+    public FeatureDataset(int newFeatures, FeatureSchema featureSchema, Envelope envelope) {
+        this.features = new ArrayList<>(newFeatures);
         this.featureSchema = featureSchema;
+        this.envelope = envelope == null ? null : envelope.copy();
     }
 
     /**
      * Creates a FeatureDataset.
      * @param featureSchema the types of the attributes of the features in this collection
      */
     public FeatureDataset(FeatureSchema featureSchema) {
-        this(new ArrayList<Feature>(), featureSchema);
+        this(new ArrayList<>(), featureSchema);
+    }
+
+    /**
+     * Clone another dataset
+     * @param otherDataset The dataset to clone
+     */
+    public FeatureDataset(FeatureDataset otherDataset) {
+        this(otherDataset.getFeatures(), otherDataset.getFeatureSchema());
+        if (otherDataset.envelope != null) {
+            this.envelope = otherDataset.envelope.copy();
+        }
     }
 
     public Feature getFeature(int index) {
         return features.get(index);
     }
 
     @Override
-	public FeatureSchema getFeatureSchema() {
+    public FeatureSchema getFeatureSchema() {
         return featureSchema;
     }
 
@@ -91,7 +115,7 @@ public FeatureSchema getFeatureSchema() {
      * later change a Feature's geometry using Feature#setGeometry.
      */
     @Override
-	public Envelope getEnvelope() {
+    public Envelope getEnvelope() {
         if (envelope == null) {
             envelope = new Envelope();
 
@@ -105,12 +129,12 @@ public Envelope getEnvelope() {
     }
 
     @Override
-	public List<Feature> getFeatures() {
+    public List<Feature> getFeatures() {
         return Collections.unmodifiableList(features);
     }
 
     @Override
-	public boolean isEmpty() {
+    public boolean isEmpty() {
         return size() == 0;
     }
 
@@ -122,7 +146,7 @@ public boolean isEmpty() {
     //<<TODO:DESIGN>> Perhaps return value should be a Set, not a List, because order
     //doesn't matter. [Jon Aquino]
     @Override
-	public List<Feature> query(Envelope envelope) {
+    public List<Feature> query(Envelope envelope) {
         if (!envelope.intersects(getEnvelope())) {
             return new ArrayList<>();
         }
@@ -141,7 +165,7 @@ public List<Feature> query(Envelope envelope) {
     }
 
     @Override
-	public void add(Feature feature) {
+    public void add(Feature feature) {
         features.add(feature);
         if (envelope != null) {
             envelope.expandToInclude(feature.getGeometry().getEnvelopeInternal());
@@ -162,15 +186,15 @@ public boolean contains(Feature feature) {
      * @param env
      */
     @Override
-	public Collection<Feature> remove(Envelope env) {
+    public Collection<Feature> remove(Envelope env) {
         Collection<Feature> features = query(env);
         removeAll(features);
 
         return features;
     }
 
     @Override
-	public void remove(Feature feature) {
+    public void remove(Feature feature) {
         features.remove(feature);
         invalidateEnvelope();
     }
@@ -179,18 +203,18 @@ public void remove(Feature feature) {
      * Removes all features from this collection.
      */
     @Override
-	public void clear() {
+    public void clear() {
         invalidateEnvelope();
         features.clear();
     }
 
     @Override
-	public int size() {
+    public int size() {
         return features.size();
     }
 
     @Override
-	public Iterator<Feature> iterator() {
+    public Iterator<Feature> iterator() {
         return features.iterator();
     }
 
@@ -199,7 +223,7 @@ public void invalidateEnvelope() {
     }
 
     @Override
-	public void addAll(Collection<? extends Feature> features) {
+    public void addAll(Collection<? extends Feature> features) {
         this.features.addAll(features);
         if (envelope != null) {
             for (Feature feature : features) {
@@ -209,7 +233,7 @@ public void addAll(Collection<? extends Feature> features) {
     }
 
     @Override
-	public void removeAll(Collection<Feature> features) {
+    public void removeAll(Collection<Feature> features) {
         this.features.removeAll(features);
         invalidateEnvelope();
     }