Put comments from the CoNLLU on the annotation

Author: AngledLuffa

src/edu/stanford/nlp/ling/CoreAnnotations.java

@@ -522,6 +522,16 @@ public Class<String> getType() {
     }
   }
 
+  /**
+   * Comments on the sentence, such as the ones attached to CoNLLU sentences
+   */
+  public static class CommentsAnnotation implements CoreAnnotation<List<String>> {
+    @Override
+    public Class<List<String>> getType() {
+      return ErasureUtils.uncheckedCast(List.class);
+    }
+  }
+
   /**
    * CoNLL dep parsing - coarser POS tags.
    */

src/edu/stanford/nlp/pipeline/CoNLLUReader.java

@@ -24,7 +24,6 @@ public class CoNLLUReader {
    * field constants
    **/
   // TODO: read sent_id?
-  // TODO: read comments in general
   // TODO: reconsider the newline as the after on the last word
   // TODO: keep around the rest of the misc annotations
   public static final int CoNLLU_IndexField = 0;
@@ -223,6 +222,8 @@ public class CoNLLUSentence {
     public List<String> emptyLines = new ArrayList<>();
     // data for the sentence contained in # key values
     public HashMap<String, String> sentenceData = new HashMap<>();
+    // all of the comments, including the ones that showed up in sentenceData
+    public List<String> comments = new ArrayList<>();
     // map indices in token list to mwt data if there is any
     HashMap<Integer, Integer> mwtData = new HashMap<>();
     // mwt tokens
@@ -259,6 +260,7 @@ public void addSentenceData(String sentenceDataLine) {
         String value = sentenceDataLine.substring(sentenceDataLine.indexOf('='));
         sentenceData.put(key, value);
       }
+      comments.add(sentenceDataLine);
     }
 
     /**
@@ -600,6 +602,7 @@ public CoreMap convertCoNLLUSentenceToCoreMap(CoNLLUDocument doc, CoNLLUSentence
       sentenceCoreMap.set(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class, enhancedParse);
     }
 
+    sentenceCoreMap.set(CoreAnnotations.CommentsAnnotation.class, sentence.comments);
     return sentenceCoreMap;
   }
 

src/edu/stanford/nlp/semgraph/semgrex/SemgrexPattern.java

@@ -580,8 +580,11 @@ public static void main(String[] args) throws IOException {
       } else if (outputFormat == OutputFormat.CONLLU) {
         CoNLLUDocumentWriter writer = new CoNLLUDocumentWriter();
         String semgrexName = semgrex.toString().trim();
-        // TODO: comments should load from the CoNLLU document, if applicable
-        List<String> comments = new ArrayList<>(graph.getComments());
+        List<String> comments = new ArrayList<>(sentence.get(CoreAnnotations.CommentsAnnotation.class));
+        // TODO: maybe stop putting comments on the graphs?
+        if (comments.size() == 0) {
+          comments.addAll(graph.getComments());
+        }
         boolean found = true;
         while (found) {
           StringBuilder comment = new StringBuilder();