wrong file displayed for unknown vector in analysis

Author: parrt

java/src/org/antlr/codebuff/CollectFeatures.java

@@ -511,7 +511,7 @@ public static int[] getNodeFeatures(Map<Token, TerminalNode> tokenToNodeMap,
 			window.get(3).getType(),
 
 			// info
-			0, // file
+			0, // dummy; we don't store file index into feature vector
 			curToken.getLine(),
 			curToken.getCharPositionInLine()
 		};

java/src/org/antlr/codebuff/FeatureType.java

@@ -2,7 +2,7 @@
 
 public enum FeatureType {
 	TOKEN(12), RULE(14), INT(7), BOOL(5), COL(7),
-	INFO_FILE(10), INFO_LINE(4), INFO_CHARPOS(4),
+	INFO_FILE(15), INFO_LINE(4), INFO_CHARPOS(4),
 	UNUSED(0);
 	public int displayWidth;
 

java/src/org/antlr/codebuff/Formatter.java

@@ -295,16 +295,16 @@ public TokenPositionAnalysis getTokenAnalysis(int[] features, int indexIntoRealT
 
 
 		String newlineAnalysis = newlinePredictionString+"\n"+
-			newlineClassifier.getPredictionAnalysis(k, features, corpus.injectNewlines,
+			newlineClassifier.getPredictionAnalysis(doc, k, features, corpus.injectNewlines,
 			                                        MAX_CONTEXT_DIFF_THRESHOLD);
 		String alignAnalysis =alignPredictionString+"\n"+
-			alignClassifier.getPredictionAnalysis(k, features, corpus.align,
+			alignClassifier.getPredictionAnalysis(doc, k, features, corpus.align,
 			                                      MAX_CONTEXT_DIFF_THRESHOLD);
 		String indentAnalysis =indentPredictionString+"\n"+
-			indentClassifier.getPredictionAnalysis(k, features, corpus.indent,
+			indentClassifier.getPredictionAnalysis(doc, k, features, corpus.indent,
 			                                       MAX_CONTEXT_DIFF_THRESHOLD);
 		String wsAnalysis =wsPredictionString+"\n"+
-			wsClassifier.getPredictionAnalysis(k, features, corpus.injectWS,
+			wsClassifier.getPredictionAnalysis(doc, k, features, corpus.injectWS,
 			                                   MAX_CONTEXT_DIFF_THRESHOLD);
 		return new TokenPositionAnalysis(newlineAnalysis, alignAnalysis, indentAnalysis, wsAnalysis);
 	}

java/src/org/antlr/codebuff/kNNClassifier.java

@@ -98,14 +98,14 @@ public HashBag<Integer> getVotesBag(Neighbor[] kNN, int k, int[] unknown, List<I
 		return votes;
 	}
 
-	public String getPredictionAnalysis(int k, int[] unknown, List<Integer> Y, double distanceThreshold) {
+	public String getPredictionAnalysis(InputDocument doc, int k, int[] unknown, List<Integer> Y, double distanceThreshold) {
 		Neighbor[] kNN = kNN(unknown, k, distanceThreshold);
 		HashBag<Integer> votes = getVotesBag(kNN, k, unknown, Y);
 
 		StringBuilder buf = new StringBuilder();
 		InputDocument firstDoc = corpus.documents.get(0); // pick any doc to get parser
 		buf.append(CollectFeatures.featureNameHeader(FEATURES));
-		buf.append(CollectFeatures._toString(FEATURES, firstDoc, unknown)+"->"+votes);
+		buf.append(CollectFeatures._toString(FEATURES, doc, unknown)+"->"+votes);
 		buf.append("\n");
 		if ( kNN.length>0 ) {
 			kNN = Arrays.copyOfRange(kNN, 0, Math.min(k, kNN.length));