add embedding based Matching

cddiff/build.gradle

@@ -15,15 +15,44 @@ dependencies {
   testImplementation "org.junit.jupiter:junit-jupiter-params:$junit_version"
   testImplementation "org.junit.vintage:junit-vintage-engine:$junit_version"
   testRuntimeOnly "org.junit.jupiter:junit-jupiter-engine:$junit_version"
-  testImplementation ("de.monticore:monticore-runtime:${project.properties['mc_version']}") capabilities {
-    requireCapability("de.monticore:monticore-runtime-tests") }
+  testImplementation("de.monticore:monticore-runtime:${project.properties['mc_version']}") capabilities {
+    requireCapability("de.monticore:monticore-runtime-tests")
+  }
   testImplementation "de.monticore:monticore-grammar-trafo:$mc_version"
 
   // dependencies for cddiff
   implementation("de.monticore.lang:od:$mc_version") {
-    exclude group: 'de.monticore.lang', module:'cd4analysis'
+    exclude group: 'de.monticore.lang', module: 'cd4analysis'
   }
   implementation "edu.mit.alloy:alloy:6"
+  implementation "com.github.vinhkhuc:jfasttext:0.5"
 }
 
+tasks.register('downloadVectors') {
+  description = "Downloads the FastText vectors, including subword vectors"
+
+  def resourceDir = "src/main/resources/"
+
+  def vectors = file("$resourceDir/crawl-300d-2M-subword.bin")
+  if (!vectors.exists()) {
+
+    def zipFile = file("$resourceDir/vectors.zip")
+    if(zipFile.exists()){
+      zipFile.delete()
+    }
+    zipFile.createNewFile()
+    println("Downloading vectors into ${zipFile.absolutePath}")
+    new URL("https://dl.fbaipublicfiles.com/fasttext/vectors-english/crawl-300d-2M-subword.zip").withInputStream { input -> zipFile.withOutputStream { it << input } }
 
+    copy {
+      from zipTree(zipFile)
+      into resourceDir
+    }
+    zipFile.delete()
+    file("$resourceDir/crawl-300d-2M-subword.vec").delete()
+    println("Vectors downloaded and extracted to $resourceDir")
+  } else {
+    println("Vectors already exist at $resourceDir")
+  }
+
+}

cddiff/src/main/java/de/monticore/cdmatcher/iterative/MutablePair.java

@@ -0,0 +1,41 @@
+/* (c) https://github.com/MontiCore/monticore */
+package de.monticore.cdmatcher.iterative;
+
+import java.util.Objects;
+
+public class MutablePair<T, V> {
+
+  private T a;
+  private V b;
+
+  public MutablePair(T a, V b) {
+    this.a = a;
+    this.b = b;
+  }
+
+  public T getA() { return a; }
+
+  public void setA(T a) { this.a = a; }
+
+  public V getB() { return b; }
+
+  public void setB(V b) { this.b = b; }
+
+  public String toString() {
+    return "(" + a + ", " + b + ")";
+  }
+
+  @Override
+  public boolean equals(Object o) {
+    if (o == null || getClass() != o.getClass())
+      return false;
+    MutablePair<?, ?> that = (MutablePair<?, ?>) o;
+    return Objects.equals(a, that.a) && Objects.equals(b, that.b);
+  }
+
+  @Override
+  public int hashCode() {
+    return Objects.hash(a, b);
+  }
+
+}

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/MatchBySimilarity.java

@@ -0,0 +1,20 @@
+/* (c) https://github.com/MontiCore/monticore */
+package de.monticore.cdmatcher.iterative.matching;
+
+import de.monticore.cdmatcher.CDSimilarity;
+import de.monticore.cdmatcher.MatchingStrategy;
+
+public class MatchBySimilarity<T> implements MatchingStrategy<T> {
+
+  private final CDSimilarity<T> similarity;
+
+  public MatchBySimilarity(CDSimilarity<T> similarity) {
+    this.similarity = similarity;
+  }
+
+  @Override
+  public double getScore(T srcElem, T tgtElem) {
+    return similarity.computeWeight(srcElem, tgtElem);
+  }
+
+}

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/association/MatchCDAssocByBestSuperType.java

@@ -3,6 +3,7 @@
 
 import de.monticore.cdassociation._ast.ASTCDAssociation;
 import de.monticore.cdbasis._ast.ASTCDType;
+import de.monticore.cdmatcher.MatchingStrategy;
 import de.monticore.cdmatcher.MultipleMatchingStrategy;
 import de.monticore.cdmatcher.iterative.matching.caching.CachedMatches;
 import de.monticore.cdmatcher.iterative.matching.caching.StructureCache;
@@ -15,18 +16,21 @@
 public class MatchCDAssocByBestSuperType extends
     MultipleMatchingStrategy<ASTCDAssociation, ASTCDType> {
 
-  public CachedMatches cachedMatches;
-  public StructureCache structureCache;
+  private final CachedMatches cachedMatches;
+  private final StructureCache structureCache;
+  private final MatchingStrategy<ASTCDAssociation> nameMatchingStrategy;
 
-  public MatchCDAssocByBestSuperType(CachedMatches cachedMatches, StructureCache structureCache) {
+  public MatchCDAssocByBestSuperType(CachedMatches cachedMatches, StructureCache structureCache,
+      MatchingStrategy<ASTCDAssociation> nameMatchingStrategy) {
     this.cachedMatches = cachedMatches;
     this.structureCache = structureCache;
+    this.nameMatchingStrategy = nameMatchingStrategy;
   }
 
   @Override
   public double getScore(ASTCDAssociation srcElem, ASTCDAssociation tgtElem) {
 
-    double nameScore = new MatchCDAssocByName().getScore(srcElem, tgtElem);
+    double nameScore = nameMatchingStrategy.getScore(srcElem, tgtElem);
     double typeScore = -1;
 
     if (structureCache.getLeftType(srcElem).isPresent() && structureCache.getLeftType(tgtElem)

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/association/MatchCDAssocByDirectType.java

@@ -13,12 +13,15 @@
 
 public class MatchCDAssocByDirectType implements MatchingStrategy<ASTCDAssociation> {
 
-  public CachedMatches cachedMatches;
-  public StructureCache structureCache;
+  private final CachedMatches cachedMatches;
+  private final StructureCache structureCache;
+  private final MatchingStrategy<ASTCDAssociation> nameMatchingStrategy;
 
-  public MatchCDAssocByDirectType(CachedMatches cachedMatches, StructureCache structureCache) {
+  public MatchCDAssocByDirectType(CachedMatches cachedMatches, StructureCache structureCache,
+      MatchingStrategy<ASTCDAssociation> nameMatchingStrategy) {
     this.cachedMatches = cachedMatches;
     this.structureCache = structureCache;
+    this.nameMatchingStrategy = nameMatchingStrategy;
   }
 
   @Override
@@ -28,7 +31,7 @@ public double getScore(ASTCDAssociation srcElem, ASTCDAssociation tgtElem) {
     Optional<ASTCDType> tgtRightType = structureCache.getRightType(tgtElem);
     Optional<ASTCDType> tgtLeftType = structureCache.getLeftType(tgtElem);
 
-    double nameScore = new MatchCDAssocByName().getScore(srcElem, tgtElem);
+    double nameScore = nameMatchingStrategy.getScore(srcElem, tgtElem);
     double typeScore = -1;
     Double leftTypeScore = null;
     Double rightTypeScore = null;

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/attribute/MatchCDAttributeByNameAndType.java

@@ -11,10 +11,13 @@
 
 public class MatchCDAttributeByNameAndType implements MatchingStrategy<ASTCDAttribute> {
 
-  public CachedMatches cachedMatches;
+  private final CachedMatches cachedMatches;
+  private final MatchingStrategy<ASTCDAttribute> nameMatchingStrategy;
 
-  public MatchCDAttributeByNameAndType(CachedMatches cachedMatches) {
+  public MatchCDAttributeByNameAndType(CachedMatches cachedMatches,
+      MatchingStrategy<ASTCDAttribute> nameMatchingStrategy) {
     this.cachedMatches = cachedMatches;
+    this.nameMatchingStrategy = nameMatchingStrategy;
   }
 
   @Override
@@ -25,7 +28,7 @@ public double getScore(ASTCDAttribute srcElem, ASTCDAttribute tgtElem) {
     Double attributeClassType = cachedMatches.getMatch(srcAttributeClassType,
         tgtAttributeClassType);
 
-    double score = new MatchCDAttributeByName().getScore(srcElem, tgtElem);
+    double score = nameMatchingStrategy.getScore(srcElem, tgtElem);
 
     if (attributeClassType != null) {
       score = mean(score, attributeClassType);

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/cdtype/MatchCDTypeByDirectAssocs.java

@@ -11,7 +11,7 @@ public class MatchCDTypeByDirectAssocs extends
     MultipleMatchingStrategy<ASTCDType, ASTCDAssociation> {
 
   private final MatchingStrategy<ASTCDAssociation> strategy;
-  public StructureCache structureCache;
+  private final StructureCache structureCache;
 
   public MatchCDTypeByDirectAssocs(MatchingStrategy<ASTCDAssociation> associationMatchingStrategy,
       StructureCache structureCache) {

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/cdtype/MatchCDTypeByDirectAttributes.java

@@ -3,26 +3,25 @@
 
 import de.monticore.cdbasis._ast.ASTCDAttribute;
 import de.monticore.cdbasis._ast.ASTCDType;
-import de.monticore.cdmatcher.iterative.matching.attribute.MatchCDAttributeByNameAndType;
+import de.monticore.cdmatcher.MatchingStrategy;
 import de.monticore.cdmatcher.MultipleMatchingStrategy;
-import de.monticore.cdmatcher.iterative.matching.caching.CachedMatches;
 import de.monticore.cdmatcher.iterative.matching.caching.StructureCache;
 
 public class MatchCDTypeByDirectAttributes extends
     MultipleMatchingStrategy<ASTCDType, ASTCDAttribute> {
 
-  public CachedMatches cachedMatches;
-  public StructureCache structureCache;
+  private final StructureCache structureCache;
+  private final MatchingStrategy<ASTCDAttribute> strategy;
 
-  public MatchCDTypeByDirectAttributes(CachedMatches cachedMatches, StructureCache structureCache) {
-    this.cachedMatches = cachedMatches;
+  public MatchCDTypeByDirectAttributes(StructureCache structureCache,
+      MatchingStrategy<ASTCDAttribute> strategy) {
     this.structureCache = structureCache;
+    this.strategy = strategy;
   }
 
   @Override
   public double getScore(ASTCDType srcElem, ASTCDType tgtElem) {
-    return getBestMatchingScore(srcElem, tgtElem, structureCache::getAttributes,
-        new MatchCDAttributeByNameAndType(cachedMatches));
+    return getBestMatchingScore(srcElem, tgtElem, structureCache::getAttributes, strategy);
   }
 
 }

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/cdtype/MatchCDTypeByDirectSubClasses.java

@@ -8,8 +8,8 @@
 
 public class MatchCDTypeByDirectSubClasses extends MultipleMatchingStrategy<ASTCDType, ASTCDType> {
 
-  public CachedMatches cachedMatches;
-  public StructureCache structureCache;
+  private final CachedMatches cachedMatches;
+  private final StructureCache structureCache;
 
   public MatchCDTypeByDirectSubClasses(CachedMatches cachedMatches, StructureCache structureCache) {
     this.cachedMatches = cachedMatches;

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/cdtype/MatchCDTypeByDirectSuperClasses.java

@@ -9,8 +9,8 @@
 public class MatchCDTypeByDirectSuperClasses extends
     MultipleMatchingStrategy<ASTCDType, ASTCDType> {
 
-  public CachedMatches cachedMatches;
-  public StructureCache structureCache;
+  private final CachedMatches cachedMatches;
+  private final StructureCache structureCache;
 
   public MatchCDTypeByDirectSuperClasses(CachedMatches cachedMatches,
       StructureCache structureCache) {

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/cdtype/MatchCDTypeComposite.java

@@ -13,8 +13,8 @@
 
 public class MatchCDTypeComposite implements MatchingStrategy<ASTCDType> {
 
-  Set<MatchingStrategy<ASTCDType>> strategies;
-  public CachedMatches cachedMatches;
+  private final Set<MatchingStrategy<ASTCDType>> strategies;
+  private final CachedMatches cachedMatches;
 
   public MatchCDTypeComposite(Set<MatchingStrategy<ASTCDType>> strategies,
       CachedMatches cachedMatches) {

cddiff/src/main/java/de/monticore/cdmatcher/iterative/matching/cdtype/MatchCDTypeFromCache.java

@@ -2,22 +2,42 @@
 package de.monticore.cdmatcher.iterative.matching.cdtype;
 
 import de.monticore.cdbasis._ast.ASTCDType;
+import de.monticore.cdmatcher.CDTypeSimilarity;
 import de.monticore.cdmatcher.MatchingStrategy;
+import de.monticore.cdmatcher.iterative.matching.MatchBySimilarity;
 import de.monticore.cdmatcher.iterative.matching.caching.CachedMatches;
 
 public class MatchCDTypeFromCache implements MatchingStrategy<ASTCDType> {
 
-  public CachedMatches cachedMatches;
+  private static MatchingStrategy<ASTCDType> defaultFallbackStrategy = new MatchBySimilarity<>(
+      new CDTypeSimilarity());
+  private final CachedMatches cachedMatches;
+  private final MatchingStrategy<ASTCDType> fallbackStrategy;
 
-  public MatchCDTypeFromCache(CachedMatches cachedMatches) {
+  public MatchCDTypeFromCache(CachedMatches cachedMatches,
+      MatchingStrategy<ASTCDType> fallbackStrategy) {
     this.cachedMatches = cachedMatches;
+    this.fallbackStrategy = fallbackStrategy;
+  }
+
+  public MatchCDTypeFromCache(CachedMatches cachedMatches) {
+    this(cachedMatches, defaultFallbackStrategy);
+  }
+
+  public static MatchingStrategy<ASTCDType> getDefaultFallbackStrategy() {
+    return defaultFallbackStrategy;
+  }
+
+  public static void setDefaultFallbackStrategy(
+      MatchingStrategy<ASTCDType> defaultFallbackStrategy) {
+    MatchCDTypeFromCache.defaultFallbackStrategy = defaultFallbackStrategy;
   }
 
   @Override
   public double getScore(ASTCDType srcElem, ASTCDType tgtElem) {
     Double cachedScore = cachedMatches.getMatch(srcElem, tgtElem);
 
-    return cachedScore != null ? cachedScore : new MatchCDTypeByName().getScore(srcElem, tgtElem);
+    return cachedScore != null ? cachedScore : fallbackStrategy.getScore(srcElem, tgtElem);
   }
 
 }