Index: lucene/analysis/common/src/java/org/apache/lucene/analysis/shingle/ShingleFilterFactory.java
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
--- lucene/analysis/common/src/java/org/apache/lucene/analysis/shingle/ShingleFilterFactory.java	(revision 1561206)
+++ lucene/analysis/common/src/java/org/apache/lucene/analysis/shingle/ShingleFilterFactory.java	(revision )
@@ -29,7 +29,7 @@
  *   &lt;analyzer&gt;
  *     &lt;tokenizer class="solr.WhitespaceTokenizerFactory"/&gt;
  *     &lt;filter class="solr.ShingleFilterFactory" minShingleSize="2" maxShingleSize="2"
- *             outputUnigrams="true" outputUnigramsIfNoShingles="false" tokenSeparator=" "/&gt;
+ *             outputUnigrams="true" outputUnigramsIfNoShingles="false" tokenSeparator=" " fillerToken="_"/&gt;
  *   &lt;/analyzer&gt;
  * &lt;/fieldType&gt;</pre>
  */
@@ -39,6 +39,7 @@
   private final boolean outputUnigrams;
   private final boolean outputUnigramsIfNoShingles;
   private final String tokenSeparator;
+  private final char [] fillerToken;
 
   /** Creates a new ShingleFilterFactory */
   public ShingleFilterFactory(Map<String, String> args) {
@@ -58,6 +59,7 @@
     outputUnigrams = getBoolean(args, "outputUnigrams", true);
     outputUnigramsIfNoShingles = getBoolean(args, "outputUnigramsIfNoShingles", false);
     tokenSeparator = get(args, "tokenSeparator", ShingleFilter.TOKEN_SEPARATOR);
+    fillerToken = null == get(args, "fillerToken") ? ShingleFilter.FILLER_TOKEN : get(args, "fillerToken").toCharArray();
     if (!args.isEmpty()) {
       throw new IllegalArgumentException("Unknown parameters: " + args);
     }
@@ -69,6 +71,7 @@
     r.setOutputUnigrams(outputUnigrams);
     r.setOutputUnigramsIfNoShingles(outputUnigramsIfNoShingles);
     r.setTokenSeparator(tokenSeparator);
+    r.setTokenFiller(fillerToken);
     return r;
   }
 }
Index: lucene/analysis/common/src/java/org/apache/lucene/analysis/shingle/ShingleFilter.java
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
--- lucene/analysis/common/src/java/org/apache/lucene/analysis/shingle/ShingleFilter.java	(revision 1561206)
+++ lucene/analysis/common/src/java/org/apache/lucene/analysis/shingle/ShingleFilter.java	(revision )
@@ -97,6 +97,8 @@
    */
   private String tokenSeparator = TOKEN_SEPARATOR;
 
+  private char[] fillerToken = FILLER_TOKEN;
+
   /**
    * By default, we output unigrams (individual tokens) as well as shingles
    * (token n-grams).
@@ -284,6 +286,15 @@
     this.tokenSeparator = null == tokenSeparator ? "" : tokenSeparator;
   }
 
+  /**
+   * Sets the string that is inserted in between the tokens if pos_inc is > 1
+   * @param fillerToken
+   */
+  public void setTokenFiller(char [] fillerToken) {
+    if(fillerToken!=null)
+      this.fillerToken = fillerToken;
+  }
+
   @Override
   public boolean incrementToken() throws IOException {
     boolean tokenAvailable = false;
@@ -341,7 +352,7 @@
   /**
    * <p>Get the next token from the input stream.
    * <p>If the next token has <code>positionIncrement > 1</code>,
-   * <code>positionIncrement - 1</code> {@link #FILLER_TOKEN}s are
+   * <code>positionIncrement - 1</code> {@link #fillerToken}s are
    * inserted first.
    * @param target Where to put the new token; if null, a new instance is created.
    * @return On success, the populated token; null otherwise
@@ -359,7 +370,7 @@
       // A filler token occupies no space
       newTarget.offsetAtt.setOffset(newTarget.offsetAtt.startOffset(), 
                                     newTarget.offsetAtt.startOffset());
-      newTarget.termAtt.copyBuffer(FILLER_TOKEN, 0, FILLER_TOKEN.length);
+      newTarget.termAtt.copyBuffer(fillerToken, 0, fillerToken.length);
       newTarget.isFiller = true;
       --numFillerTokensToInsert;
     } else if (isNextInputStreamToken) {
@@ -390,7 +401,7 @@
           isNextInputStreamToken = true;
           // A filler token occupies no space
           newTarget.offsetAtt.setOffset(offsetAtt.startOffset(), offsetAtt.startOffset());
-          newTarget.termAtt.copyBuffer(FILLER_TOKEN, 0, FILLER_TOKEN.length);
+          newTarget.termAtt.copyBuffer(fillerToken, 0, fillerToken.length);
           newTarget.isFiller = true;
           --numFillerTokensToInsert;
         } else {
Index: lucene/analysis/common/src/test/org/apache/lucene/analysis/shingle/ShingleFilterTest.java
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
--- lucene/analysis/common/src/test/org/apache/lucene/analysis/shingle/ShingleFilterTest.java	(revision 1561206)
+++ lucene/analysis/common/src/test/org/apache/lucene/analysis/shingle/ShingleFilterTest.java	(revision )
@@ -1196,4 +1196,20 @@
                               new int[] {1, 0, 0, 1, 0, 0},
                               20);
   }
+
+  public void testTwoTrailingHolesTriShingleWithTokenFiller() throws IOException {
+    // Analyzing "purple wizard of the", where of and the are removed as a
+    // stopwords, leaving two trailing holes:
+    Token[] inputTokens = new Token[] {createToken("purple", 0, 6),
+        createToken("wizard", 7, 13)};
+    ShingleFilter filter = new ShingleFilter(new CannedTokenStream(2, 20, inputTokens), 2, 3);
+    filter.setTokenFiller("--".toCharArray());
+
+    assertTokenStreamContents(filter,
+        new String[] {"purple", "purple wizard", "purple wizard --", "wizard", "wizard --", "wizard -- --"},
+        new int[] {0, 0, 0, 7, 7, 7},
+        new int[] {6, 13, 20, 13, 20, 20},
+        new int[] {1, 0, 0, 1, 0, 0},
+        20);
+  }
 }