Merge pull request #1154 from stanfordnlp/english_enhanced

UniversalEnhancer can process English separately from the others, sin…

Author: AngledLuffa

src/edu/stanford/nlp/trees/UniversalEnglishGrammaticalStructure.java

@@ -896,11 +896,10 @@ private static void convertRel(SemanticGraph sg) {
     }
   }
 
-  @Override
-  protected void addEnhancements(List<TypedDependency> list, EnhancementOptions options) {
-
-    SemanticGraph sg = new SemanticGraph(list);
-
+  /**
+   * Manipulates the given SemanticGraph to add enhancements.  Note that this modifies the input graph.
+   */
+  public static void addEnhancements(SemanticGraph sg, EnhancementOptions options) {
     if (DEBUG) {
       printListSorted("addEnhancements: before correctDependencies()", sg.typedDependencies());
     }
@@ -977,6 +976,15 @@ protected void addEnhancements(List<TypedDependency> list, EnhancementOptions op
     }
 
     correctSubjPass(sg);
+  }
+  
+  @Override
+  protected void addEnhancements(List<TypedDependency> list, EnhancementOptions options) {
+
+    SemanticGraph sg = new SemanticGraph(list);
+
+    addEnhancements(sg, options);
+
     list.clear();
     list.addAll(sg.typedDependencies());
 

src/edu/stanford/nlp/trees/ud/ProcessUniversalEnhancerRequest.java

@@ -12,6 +12,7 @@
 import edu.stanford.nlp.semgraph.SemanticGraph;
 import edu.stanford.nlp.semgraph.SemanticGraphCoreAnnotations;
 import edu.stanford.nlp.trees.EnglishPatterns;
+import edu.stanford.nlp.trees.UniversalEnglishGrammaticalStructure;
 import edu.stanford.nlp.util.CoreMap;
 import edu.stanford.nlp.util.ProcessProtobufRequest;
 
@@ -29,14 +30,36 @@ public static void enhanceDependencies(Pattern relativePronounsPattern, Annotati
     }
   }
 
+  /**
+   * Enhance the dependencies on a single sentence, using the English-specific rules for enhancement
+   */
+  public static void enhanceEnglishDependencies(Annotation annotation) {
+    for (CoreMap sentence : annotation.get(CoreAnnotations.SentencesAnnotation.class)) {
+      SemanticGraph basic = sentence.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
+
+      SemanticGraph enhanced = new SemanticGraph(basic);
+      UniversalEnglishGrammaticalStructure.addEnhancements(enhanced, UniversalEnglishGrammaticalStructure.ENHANCED_OPTIONS);
+      sentence.set(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class, enhanced);
+
+      SemanticGraph plusplus = new SemanticGraph(basic);
+      UniversalEnglishGrammaticalStructure.addEnhancements(plusplus, UniversalEnglishGrammaticalStructure.ENHANCED_PLUS_PLUS_OPTIONS);
+      sentence.set(SemanticGraphCoreAnnotations.EnhancedPlusPlusDependenciesAnnotation.class, plusplus);
+    }
+  }
+
   /**
    * Process all sentences in the document, enhancing the basic dependencies on each sentence
    */
   public static CoreNLPProtos.Document processRequest(Pattern relativePronounsPattern, CoreNLPProtos.DependencyEnhancerRequest request) {
     ProtobufAnnotationSerializer serializer = new ProtobufAnnotationSerializer();
     Annotation annotation = serializer.fromProto(request.getDocument());
 
-    enhanceDependencies(relativePronounsPattern, annotation);
+    if (request.hasLanguage() &&
+        (request.getLanguage() == CoreNLPProtos.Language.English || request.getLanguage() == CoreNLPProtos.Language.UniversalEnglish)) {
+      enhanceEnglishDependencies(annotation);
+    } else {
+      enhanceDependencies(relativePronounsPattern, annotation);
+    }
 
     return serializer.toProto(annotation);
   }