LUCENE-9093: UnifiedHighlighter LengthGoalBreakIterator frag align

Matches in passages should be centered better on average.
 Closes #1123

lucene/CHANGES.txt

@@ -83,7 +83,10 @@ Other
 
 API Changes
 ---------------------
-(No changes)
+
+* LUCENE-9093: Not an API change but a change in behavior of the UnifiedHighlighter's LengthGoalBreakIterator that will
+  yield Passages sized a little different due to the fact that the sizing pivot is now the center of the first match and
+  not its left edge.
 
 New Features
 ---------------------
@@ -106,6 +109,10 @@ Improvements
 
 * LUCENE-9106: UniformSplit postings format allows extension of block/line serializers. (Bruno Roustant)
 
+* LUCENE-9093: UnifiedHighlighter's LengthGoalBreakIterator has a new fragmentAlignment option to better center the
+  first match in the passage. Also the sizing point now pivots at the center of the first match term and not its left
+  edge. This yields Passages that won't be identical to the previous behavior. (Nándor Mátravölgyi, David Smiley)
+
 Optimizations
 ---------------------
 (No changes)

lucene/highlighter/src/java/org/apache/lucene/search/uhighlight/FieldHighlighter.java

@@ -141,6 +141,7 @@ public class FieldHighlighter {
       }
     });
     Passage passage = new Passage(); // the current passage in-progress.  Will either get reset or added to queue.
+    int lastPassageEnd = 0;
 
     do {
       int start = off.startOffset();
@@ -158,9 +159,12 @@ public class FieldHighlighter {
         if (start >= contentLength) {
           break;
         }
+        // find fragment from the middle of the match, so the result's length may be closer to fragsize
+        final int center = start + (end - start) / 2;
         // advance breakIterator
-        passage.setStartOffset(Math.max(this.breakIterator.preceding(start + 1), 0));
+        passage.setStartOffset(Math.min(start, Math.max(this.breakIterator.preceding(Math.max(start + 1, center)), lastPassageEnd)));
-        passage.setEndOffset(Math.min(this.breakIterator.following(start), contentLength));
+        lastPassageEnd = Math.max(end, Math.min(this.breakIterator.following(Math.min(end - 1, center)), contentLength));
+        passage.setEndOffset(lastPassageEnd);
       }
       // Add this term to the passage.
       BytesRef term = off.getTerm();// a reference; safe to refer to

lucene/highlighter/src/java/org/apache/lucene/search/uhighlight/LengthGoalBreakIterator.java

@@ -33,22 +33,49 @@ public class LengthGoalBreakIterator extends BreakIterator {
 
   private final BreakIterator baseIter;
   private final int lengthGoal;
+  private final float fragmentAlignment; // how much text to align before match-fragment, valid in range [0, 1]
   private final boolean isMinimumLength; // if false then is "closest to" length
+  private int currentCache;
 
-  /** Breaks will be at least {@code minLength} apart (to the extent possible). */
+  /**
+   * Breaks will be at least {@code minLength} apart (to the extent possible),
+   * while trying to position the match inside the fragment according to {@code fragmentAlignment}.
+   */
+  public static LengthGoalBreakIterator createMinLength(BreakIterator baseIter, int minLength,
+                                                        float fragmentAlignment) {
+    return new LengthGoalBreakIterator(baseIter, minLength, fragmentAlignment, true, baseIter.current());
+  }
+
+  /** For backwards compatibility you can initialise the break iterator without fragmentAlignment. */
+  @Deprecated
   public static LengthGoalBreakIterator createMinLength(BreakIterator baseIter, int minLength) {
-    return new LengthGoalBreakIterator(baseIter, minLength, true);
+    return createMinLength(baseIter, minLength, 0.f);
   }
 
-  /** Breaks will be on average {@code targetLength} apart; the closest break to this target (before or after)
+  /**
-   * is chosen. */
+   * Breaks will be on average {@code targetLength} apart; the closest break to this target (before or after)
+   * is chosen. The match will be positioned according to {@code fragmentAlignment} as much as possible.
+   */
+  public static LengthGoalBreakIterator createClosestToLength(BreakIterator baseIter, int targetLength,
+                                                              float fragmentAlignment) {
+    return new LengthGoalBreakIterator(baseIter, targetLength, fragmentAlignment, false, baseIter.current());
+  }
+
+  /** For backwards compatibility you can initialise the break iterator without fragmentAlignment. */
+  @Deprecated
   public static LengthGoalBreakIterator createClosestToLength(BreakIterator baseIter, int targetLength) {
-    return new LengthGoalBreakIterator(baseIter, targetLength, false);
+    return createClosestToLength(baseIter, targetLength, 0.f);
   }
 
-  private LengthGoalBreakIterator(BreakIterator baseIter, int lengthGoal, boolean isMinimumLength) {
+  private LengthGoalBreakIterator(BreakIterator baseIter, int lengthGoal, float fragmentAlignment,
+                                  boolean isMinimumLength, int currentCache) {
     this.baseIter = baseIter;
+    this.currentCache = currentCache;
     this.lengthGoal = lengthGoal;
+    if (fragmentAlignment < 0.f || fragmentAlignment > 1.f || !Float.isFinite(fragmentAlignment)) {
+      throw new IllegalArgumentException("fragmentAlignment must be >= zero and <= one");
+    }
+    this.fragmentAlignment = fragmentAlignment;
     this.isMinimumLength = isMinimumLength;
   }
 
@@ -60,12 +87,15 @@ public class LengthGoalBreakIterator extends BreakIterator {
   @Override
   public String toString() {
     String goalDesc = isMinimumLength ? "minLen" : "targetLen";
-    return getClass().getSimpleName() + "{" + goalDesc + "=" + lengthGoal + ", baseIter=" + baseIter + "}";
+    return getClass().getSimpleName() + "{" + goalDesc + "=" + lengthGoal + ", fragAlign=" + fragmentAlignment +
+        ", baseIter=" + baseIter + "}";
   }
 
   @Override
   public Object clone() {
-    return new LengthGoalBreakIterator((BreakIterator) baseIter.clone(), lengthGoal, isMinimumLength);
+    return new LengthGoalBreakIterator(
+        (BreakIterator) baseIter.clone(), lengthGoal, fragmentAlignment, isMinimumLength, currentCache
+    );
   }
 
   @Override
@@ -76,26 +106,28 @@ public class LengthGoalBreakIterator extends BreakIterator {
   @Override
   public void setText(String newText) {
     baseIter.setText(newText);
+    currentCache = baseIter.current();
   }
 
   @Override
   public void setText(CharacterIterator newText) {
     baseIter.setText(newText);
+    currentCache = baseIter.current();
   }
 
   @Override
   public int current() {
-    return baseIter.current();
+    return currentCache;
   }
 
   @Override
   public int first() {
-    return baseIter.first();
+    return currentCache = baseIter.first();
   }
 
   @Override
   public int last() {
-    return baseIter.last();
+    return currentCache = baseIter.last();
   }
 
   @Override
@@ -104,10 +136,10 @@ public class LengthGoalBreakIterator extends BreakIterator {
     return baseIter.next(n); // probably wrong
   }
 
-  // called by getSummaryPassagesNoHighlight to generate default summary.
+  // Called by getSummaryPassagesNoHighlight to generate default summary.
   @Override
   public int next() {
-    return following(current());
+    return following(currentCache, currentCache + lengthGoal);
   }
 
   @Override
@@ -116,65 +148,70 @@ public class LengthGoalBreakIterator extends BreakIterator {
     return baseIter.previous();
   }
 
-  // called while the current position is the start of a new passage; find end of passage
   @Override
-  public int following(int followingIdx) {
+  public int following(int matchEndIndex) {
-    final int startIdx = current();
+    return following(matchEndIndex, (matchEndIndex + 1) + (int)(lengthGoal * (1.f - fragmentAlignment)));
-    if (followingIdx < startIdx) {
-      assert false : "Not supported";
-      return baseIter.following(followingIdx);
-    }
-    final int targetIdx = startIdx + lengthGoal;
-    // When followingIdx >= targetIdx, we can simply delegate since it will be >= the target
-    if (followingIdx >= targetIdx - 1) {
-      return baseIter.following(followingIdx);
-    }
-    // If target exceeds the text length, return the last index.
-    if (targetIdx >= getText().getEndIndex()) {
-      return baseIter.last();
-    }
-
-    // Find closest break >= the target
-    final int afterIdx = baseIter.following(targetIdx - 1);
-    if (afterIdx == DONE) { // we're at the end; can this happen?
-      return current();
-    }
-    if (afterIdx == targetIdx) { // right on the money
-      return afterIdx;
-    }
-    if (isMinimumLength) { // thus never undershoot
-      return afterIdx;
-    }
-
-    // note: it is a shame that we invoke preceding() *in addition to* following(); BI's are sometimes expensive.
-
-    // Find closest break < target
-    final int beforeIdx = baseIter.preceding(targetIdx); // or could do baseIter.previous() but we hope the BI implements preceding()
-    if (beforeIdx <= followingIdx) { // too far back
-      return moveToBreak(afterIdx);
-    }
-
-    if (targetIdx - beforeIdx <= afterIdx - targetIdx) {
-      return beforeIdx;
-    }
-    return moveToBreak(afterIdx);
   }
 
-  private int moveToBreak(int idx) { // precondition: idx is a known break
+  private int following(int matchEndIndex, int targetIdx) {
-    // bi.isBoundary(idx) has side-effect of moving the position.  Not obvious!
+    if (targetIdx >= getText().getEndIndex()) {
-    //boolean moved = baseIter.isBoundary(idx); // probably not particularly expensive
+      if (currentCache == baseIter.last()) {
-    //assert moved && current() == idx;
+        return DONE;
+      }
+      return currentCache = baseIter.last();
+    }
+    final int afterIdx = baseIter.following(targetIdx - 1);
+    if (afterIdx == DONE) {
+      currentCache = baseIter.last();
+      return DONE;
+    }
+    if (afterIdx == targetIdx) { // right on the money
+      return currentCache = afterIdx;
+    }
+    if (isMinimumLength) { // thus never undershoot
+      return currentCache = afterIdx;
+    }
 
-    // TODO fix: Would prefer to do "- 1" instead of "- 2" but CustomSeparatorBreakIterator has a bug.
+    // note: it is a shame that we invoke preceding() *one more time*; BI's are sometimes expensive.
-    int current = baseIter.following(idx - 2);
+
-    assert current == idx : "following() didn't move us to the expected index.";
+    // Find closest break to target
-    return idx;
+    final int beforeIdx = baseIter.preceding(targetIdx);
+    if (targetIdx - beforeIdx < afterIdx - targetIdx && beforeIdx > matchEndIndex) {
+      return currentCache = beforeIdx;
+    }
+    return currentCache = afterIdx;
   }
 
   // called at start of new Passage given first word start offset
   @Override
-  public int preceding(int offset) {
+  public int preceding(int matchStartIndex) {
-    return baseIter.preceding(offset); // no change needed
+    final int targetIdx = (matchStartIndex - 1) - (int)(lengthGoal * fragmentAlignment);
+    if (targetIdx <= 0) {
+      if (currentCache == baseIter.first()) {
+        return DONE;
+      }
+      return currentCache = baseIter.first();
+    }
+    final int beforeIdx = baseIter.preceding(targetIdx + 1);
+    if (beforeIdx == DONE) {
+      currentCache = baseIter.first();
+      return DONE;
+    }
+    if (beforeIdx == targetIdx) { // right on the money
+      return currentCache = beforeIdx;
+    }
+    if (isMinimumLength) { // thus never undershoot
+      return currentCache = beforeIdx;
+    }
+
+    // note: it is a shame that we invoke following() *one more time*; BI's are sometimes expensive.
+
+    // Find closest break to target
+    final int afterIdx = baseIter.following(targetIdx - 1);
+    if (afterIdx - targetIdx < targetIdx - beforeIdx && afterIdx < matchStartIndex) {
+      return currentCache = afterIdx;
+    }
+    return currentCache = beforeIdx;
   }
 
   @Override

lucene/highlighter/src/test/org/apache/lucene/search/uhighlight/LengthGoalBreakIteratorTest.java

@@ -18,6 +18,7 @@
 package org.apache.lucene.search.uhighlight;
 
 import java.io.IOException;
+import java.text.BreakIterator;
 
 import org.apache.lucene.analysis.Analyzer;
 import org.apache.lucene.analysis.MockAnalyzer;
@@ -28,6 +29,7 @@ import org.apache.lucene.util.QueryBuilder;
 
 public class LengthGoalBreakIteratorTest extends LuceneTestCase {
   private static final String FIELD = "body";
+  private static final float[] ALIGNS = {0.f, 0.5f, 1.f};
 
   // We test LengthGoalBreakIterator as it is used by the UnifiedHighlighter instead of directly, because it is
   //  not a general purpose BreakIterator.  A unit test of it directly wouldn't give as much confidence.
@@ -39,65 +41,159 @@ public class LengthGoalBreakIteratorTest extends LuceneTestCase {
   //                      0         1
   //                      01234567890123456789
   static final String CONTENT = "Aa bb. Cc dd. Ee ff";
+  static final String CONTENT2 = "Aa bb Cc dd X Ee ff Gg hh.";
+  static final String CONTENT3 = "Aa bbcc ddxyzee ffgg hh.";
+
+  public void testFragmentAlignmentConstructor() throws IOException {
+    BreakIterator baseBI = new CustomSeparatorBreakIterator('.');
+    // test fragmentAlignment validation
+    float[] valid_aligns = {0.f, 0.3333f, 0.5f, 0.99f, 1.f};
+    for (float alignment : valid_aligns) {
+      LengthGoalBreakIterator.createClosestToLength(baseBI, 50, alignment);
+    }
+    float[] invalid_aligns = {-0.01f, -1.f, 1.5f, Float.NaN, Float.NEGATIVE_INFINITY, Float.POSITIVE_INFINITY};
+    for (float alignment : invalid_aligns) {
+      expectThrows(IllegalArgumentException.class, () -> {
+        LengthGoalBreakIterator.createClosestToLength(baseBI, 50, alignment);
+      });
+    }
+    // test backwards compatibility constructors
+    String backwardCompString = LengthGoalBreakIterator.createClosestToLength(baseBI, 50).toString();
+    assertTrue(backwardCompString, backwardCompString.contains("fragAlign=0.0"));
+    backwardCompString = LengthGoalBreakIterator.createMinLength(baseBI, 50).toString();
+    assertTrue(backwardCompString, backwardCompString.contains("fragAlign=0.0"));
+  }
 
   public void testTargetLen() throws IOException {
     // "goal" means target length goal to find closest break
 
     // at first word:
     Query query = query("aa");
-    assertEquals("almost two sent",
+    assertEquals("almost two sent A",
-        "<b>Aa</b> bb.", highlightClosestToLen(CONTENT, query, 9));
+        "<b>Aa</b> bb.", highlightClosestToLen(CONTENT, query, 7, 0.f));
-    assertEquals( "barely two sent",
+    assertEquals("almost two sent B",
-        "<b>Aa</b> bb. Cc dd.", highlightClosestToLen(CONTENT, query, 10));
+        "<b>Aa</b> bb.", highlightClosestToLen(CONTENT, query, 15, 0.5f));
-    assertEquals("long goal",
+    assertEquals("almost two sent C",
-        "<b>Aa</b> bb. Cc dd. Ee ff", highlightClosestToLen(CONTENT, query, 17 + random().nextInt(20)));
+        "<b>Aa</b> bb.", highlightClosestToLen(CONTENT, query, 64, 1.f));
-
+    assertEquals("barely two sent A",
+        "<b>Aa</b> bb. Cc dd.", highlightClosestToLen(CONTENT, query, 8, 0.f));
+    assertEquals("barely two sent B",
+        "<b>Aa</b> bb. Cc dd.", highlightClosestToLen(CONTENT, query, 16, 0.5f));
+    assertEquals("long goal A",
+        "<b>Aa</b> bb. Cc dd. Ee ff", highlightClosestToLen(CONTENT, query, 14 + random().nextInt(20), 0.f));
+    assertEquals("long goal B",
+        "<b>Aa</b> bb. Cc dd. Ee ff", highlightClosestToLen(CONTENT, query, 28 + random().nextInt(20), 0.5f));
     // at some word not at start of passage
     query = query("dd");
-    assertEquals("short goal",
+    for (float align : ALIGNS) {
-        " Cc <b>dd</b>.", highlightClosestToLen(CONTENT, query, random().nextInt(5)));
+      // alignment is not meaningful if fragsize is shorter than or closer to match-fragment boundaries
-    assertEquals("almost two sent",
+      assertEquals("short goal " + align,
-        " Cc <b>dd</b>.", highlightClosestToLen(CONTENT, query, 10));
+          " Cc <b>dd</b>.", highlightClosestToLen(CONTENT, query, random().nextInt(4), align));
-    assertEquals("barely two sent",
+    }
-        " Cc <b>dd</b>. Ee ff", highlightClosestToLen(CONTENT, query, 11));
+    // preceding/following inclusion by alignment parameter
-    assertEquals("long goal",
+    assertEquals("barely two sent A",
-        " Cc <b>dd</b>. Ee ff", highlightClosestToLen(CONTENT, query, 12 + random().nextInt(20)));
+        " Cc <b>dd</b>. Ee ff", highlightClosestToLen(CONTENT, query, 11, 0.f));
+    assertEquals("barely two sent B",
+        " Cc <b>dd</b>. Ee ff", highlightClosestToLen(CONTENT, query, 11, 0.5f));
+    assertEquals("barely two sent C",
+        "Aa bb. Cc <b>dd</b>.", highlightClosestToLen(CONTENT, query, 11, 1.f));
+    assertEquals("long goal A",
+        " Cc <b>dd</b>. Ee ff", highlightClosestToLen(CONTENT, query, 17 + random().nextInt(20), 0.f));
+    assertEquals("long goal B",
+        "Aa bb. Cc <b>dd</b>. Ee ff", highlightClosestToLen(CONTENT, query, 17 + random().nextInt(20), 0.5f));
+    assertEquals("long goal C",
+        "Aa bb. Cc <b>dd</b>.", highlightClosestToLen(CONTENT, query, 17 + random().nextInt(20), 1.f));
+
+    query = query("ddxyzee");
+    assertEquals("test fragment search from the middle of the match; almost including",
+        "<b>ddxyzee</b> ", highlightClosestToLen(CONTENT3, query, 7, 0.5f, 1, ' '));
+    assertEquals("test fragment search from the middle of the match; barely including",
+        "bbcc <b>ddxyzee</b> ffgg ", highlightClosestToLen(CONTENT3, query, 14, 0.5f, 1, ' '));
   }
 
   public void testMinLen() throws IOException {
     // minLen mode is simpler than targetLen... just test a few cases
 
     Query query = query("dd");
-    assertEquals("almost two sent",
+    assertEquals("almost two sent A",
-        " Cc <b>dd</b>.", highlightMinLen(CONTENT, query, 6));
+        " Cc <b>dd</b>.", highlightMinLen(CONTENT, query, 0, 0.f));
-    assertEquals("barely two sent",
+    assertEquals("almost two sent B",
-        " Cc <b>dd</b>. Ee ff", highlightMinLen(CONTENT, query, 7));
+        " Cc <b>dd</b>.", highlightMinLen(CONTENT, query, 1, 0.5f));
+    assertEquals("almost two sent C",
+        " Cc <b>dd</b>.", highlightMinLen(CONTENT, query, 5, 1.f));
+
+    assertEquals("barely two sent A",
+        " Cc <b>dd</b>. Ee ff", highlightMinLen(CONTENT, query, 1, 0.f));
+    assertEquals("barely two sent B",
+        " Cc <b>dd</b>. Ee ff", highlightMinLen(CONTENT, query, 2, 0.5f));
+    assertEquals("barely two sent C",
+        "Aa bb. Cc <b>dd</b>.", highlightMinLen(CONTENT, query, 7, 1.f));
+    assertEquals("barely two sent D/a",
+        " Cc <b>dd</b>.", highlightMinLen(CONTENT, query, 2, 0.55f));
+    assertEquals("barely two sent D/b",
+        " Cc <b>dd</b>. Ee ff", highlightMinLen(CONTENT, query, 3, 0.55f));
+    assertEquals("barely two sent E/a",
+        " Cc <b>dd</b>. Ee ff", highlightMinLen(CONTENT, query, 10, 0.5f));
+    assertEquals("barely two sent E/b",
+        "Aa bb. Cc <b>dd</b>. Ee ff", highlightMinLen(CONTENT, query, 10, 0.7f));
+    assertEquals("barely two sent E/c",
+        "Aa bb. Cc <b>dd</b>.", highlightMinLen(CONTENT, query, 9, 0.9f));
+
+    query = query("ddxyzee");
+    assertEquals("test fragment search from the middle of the match; almost including",
+        "<b>ddxyzee</b> ", highlightMinLen(CONTENT3, query, 7, 0.5f, ' '));
+    assertEquals("test fragment search from the middle of the match; barely including",
+        "bbcc <b>ddxyzee</b> ffgg ", highlightMinLen(CONTENT3, query, 8, 0.5f, ' '));
+  }
+
+  public void testMinLenPrecision() throws IOException {
+    Query query = query("x");
+    assertEquals("test absolute minimal length",
+        "<b>X</b> ", highlightMinLen(CONTENT2, query, 1, 0.5f, ' '));
+    assertEquals("test slightly above minimal length",
+        "dd <b>X</b> Ee ", highlightMinLen(CONTENT2, query, 4, 0.5f, ' '));
   }
 
   public void testDefaultSummaryTargetLen() throws IOException {
     Query query = query("zz");
-    assertEquals("Aa bb.",
+    for (float align : ALIGNS) { // alignment is not used for creating default-summary
-        highlightClosestToLen(CONTENT, query, random().nextInt(10))); // < 10
+      assertEquals("Aa bb.",
+          highlightClosestToLen(CONTENT, query, 6 + random().nextInt(4), align));
+      assertEquals("Aa bb. Cc dd.",
+          highlightClosestToLen(CONTENT, query, 12 + random().nextInt(4), align));
+      assertEquals("Aa bb. Cc dd. Ee ff",
+          highlightClosestToLen(CONTENT, query, 17 + random().nextInt(20), align));
+    }
     assertEquals("Aa bb. Cc dd.",
-        highlightClosestToLen(CONTENT, query, 10 + 6)); // cusp of adding 3rd sentence
+        highlightClosestToLen(CONTENT, query, 6 + random().nextInt(4), 0.f, 2));
-    assertEquals("Aa bb. Cc dd. Ee ff",
-        highlightClosestToLen(CONTENT, query, 17 + random().nextInt(20))); // >= 14
   }
 
   private Query query(String qStr) {
     return new QueryBuilder(analyzer).createBooleanQuery(FIELD, qStr);
   }
 
-  private String highlightClosestToLen(String content, Query query, int lengthGoal) throws IOException {
+  private String highlightClosestToLen(String content, Query query, int lengthGoal, float fragAlign) throws IOException {
-    UnifiedHighlighter highlighter = new UnifiedHighlighter(null, analyzer);
+    return highlightClosestToLen(content, query, lengthGoal, fragAlign, 1);
-    highlighter.setBreakIterator(() -> LengthGoalBreakIterator.createClosestToLength(new CustomSeparatorBreakIterator('.'), lengthGoal));
-    return highlighter.highlightWithoutSearcher(FIELD, query, content, 1).toString();
   }
 
-  private String highlightMinLen(String content, Query query, int lengthGoal) throws IOException {
+  private String highlightClosestToLen(String content, Query query, int lengthGoal, float fragAlign, int maxPassages) throws IOException {
+    return highlightClosestToLen(content, query, lengthGoal, fragAlign, maxPassages, '.');
+  }
+
+  private String highlightClosestToLen(String content, Query query, int lengthGoal, float fragAlign, int maxPassages, char separator) throws IOException {
+    UnifiedHighlighter highlighter = new UnifiedHighlighter(null, analyzer);
+    highlighter.setBreakIterator(() -> LengthGoalBreakIterator.createClosestToLength(new CustomSeparatorBreakIterator(separator), lengthGoal, fragAlign));
+    return highlighter.highlightWithoutSearcher(FIELD, query, content, maxPassages).toString();
+  }
+
+  private String highlightMinLen(String content, Query query, int lengthGoal, float fragAlign) throws IOException {
+    return highlightMinLen(content, query, lengthGoal, fragAlign, '.');
+  }
+
+  private String highlightMinLen(String content, Query query, int lengthGoal, float fragAlign, char separator) throws IOException {
     // differs from above only by "createMinLength"
     UnifiedHighlighter highlighter = new UnifiedHighlighter(null, analyzer);
-    highlighter.setBreakIterator(() -> LengthGoalBreakIterator.createMinLength(new CustomSeparatorBreakIterator('.'), lengthGoal));
+    highlighter.setBreakIterator(() -> LengthGoalBreakIterator.createMinLength(new CustomSeparatorBreakIterator(separator), lengthGoal, fragAlign));
     return highlighter.highlightWithoutSearcher(FIELD, query, content, 1).toString();
   }
 }

solr/CHANGES.txt

@@ -184,6 +184,10 @@ Improvements
 
 * SOLR-14129: Reuse Jackson ObjectMapper in AuditLoggerPlugin (janhoy)
 
+* LUCENE-9093: The Unified highlighter has two new passage sizing parameters, hl.fragAlignRatio and
+  hl.fragsizeIsMinimum, with defaults that aim to better center matches in fragments than previously. See the ref guide.
+  Regardless of the settings, the passages may be sized differently than before. (Nándor Mátravölgyi, David Smiley)
+
 Optimizations
 ---------------------
 (No changes)

solr/core/src/java/org/apache/solr/highlight/UnifiedSolrHighlighter.java

@@ -326,8 +326,11 @@ public class UnifiedSolrHighlighter extends SolrHighlighter implements PluginInf
       if (fragsize <= 1) { // no real minimum size
         return baseBI;
       }
-      return LengthGoalBreakIterator.createMinLength(baseBI, fragsize);
+      float fragalign = params.getFieldFloat(field, HighlightParams.FRAGALIGNRATIO, 0.5f);
-      // TODO option for using createClosestToLength()
+      if (params.getFieldBool(field, HighlightParams.FRAGSIZEISMINIMUM, false)) {
+        return LengthGoalBreakIterator.createMinLength(baseBI, fragsize, fragalign);
+      }
+      return LengthGoalBreakIterator.createClosestToLength(baseBI, fragsize, fragalign);
     }
 
     /**

solr/core/src/test/org/apache/solr/highlight/TestUnifiedSolrHighlighter.java

@@ -244,15 +244,24 @@ public class TestUnifiedSolrHighlighter extends SolrTestCaseJ4 {
     assertU(adoc("text", "This document contains # special characters, while the other document contains the same # special character.", "id", "103"));
     assertU(adoc("text", "While the other document contains the same # special character.", "id", "104"));
     assertU(commit());
-    assertQ("CUSTOM breakiterator", 
+    assertQ("CUSTOM breakiterator",
         req("q", "text:document", "sort", "id asc", "hl", "true", "hl.bs.type", "SEPARATOR","hl.bs.separator","#","hl.fragsize", "-1"),
         "//lst[@name='highlighting']/lst[@name='103']/arr[@name='text']/str='This <em>document</em> contains #'");
-    assertQ("different breakiterator", 
+    assertQ("different breakiterator",
         req("q", "text:document", "sort", "id asc", "hl", "true", "hl.bs.type", "SEPARATOR","hl.bs.separator","#","hl.fragsize", "-1"),
         "//lst[@name='highlighting']/lst[@name='104']/arr[@name='text']/str='While the other <em>document</em> contains the same #'");
 
+    assertQ("CUSTOM breakiterator with fragsize 70 minimum",
+        req("q", "text:document", "sort", "id asc", "hl", "true", "hl.bs.type", "SEPARATOR","hl.bs.separator","#","hl.fragsize", "70", "hl.fragsizeIsMinimum", "true"),
+        "//lst[@name='highlighting']/lst[@name='103']/arr[@name='text']/str='This <em>document</em> contains # special characters, while the other <em>document</em> contains the same #'");
     assertQ("CUSTOM breakiterator with fragsize 70",
         req("q", "text:document", "sort", "id asc", "hl", "true", "hl.bs.type", "SEPARATOR","hl.bs.separator","#","hl.fragsize", "70"),
+        "//lst[@name='highlighting']/lst[@name='103']/arr[@name='text']/str='This <em>document</em> contains #'");
+    assertQ("CUSTOM breakiterator with fragsize 90",
+        req("q", "text:document", "sort", "id asc", "hl", "true", "hl.bs.type", "SEPARATOR","hl.bs.separator","#","hl.fragsize", "90"),
+        "//lst[@name='highlighting']/lst[@name='103']/arr[@name='text']/str='This <em>document</em> contains #'");
+    assertQ("CUSTOM breakiterator with fragsize 100",
+        req("q", "text:document", "sort", "id asc", "hl", "true", "hl.bs.type", "SEPARATOR","hl.bs.separator","#","hl.fragsize", "100"),
         "//lst[@name='highlighting']/lst[@name='103']/arr[@name='text']/str='This <em>document</em> contains # special characters, while the other <em>document</em> contains the same #'");
   }
 
@@ -262,11 +271,17 @@ public class TestUnifiedSolrHighlighter extends SolrTestCaseJ4 {
     assertU(adoc("id", "10", "text", "This is a sentence just under seventy chars in length blah blah. Next sentence is here."));
     assertU(commit());
     assertQ("default fragsize",
-        req("q", "text:seventy", "hl", "true"),
+        req("q", "text:seventy", "hl", "true", "hl.fragsizeIsMinimum", "true"),
         "//lst[@name='highlighting']/lst[@name='10']/arr[@name='text']/str='This is a sentence just under <em>seventy</em> chars in length blah blah. Next sentence is here.'");
-    assertQ("smaller fragsize",
+    assertQ("default fragsize",
-        req("q", "text:seventy", "hl", "true", "hl.fragsize", "60"), // a bit smaller
+        req("q", "text:seventy", "hl", "true", "hl.fragsizeIsMinimum", "true", "hl.fragsize", "60"),
         "//lst[@name='highlighting']/lst[@name='10']/arr[@name='text']/str='This is a sentence just under <em>seventy</em> chars in length blah blah. '");
+    assertQ("smaller fragsize",
+        req("q", "text:seventy", "hl", "true"),
+        "//lst[@name='highlighting']/lst[@name='10']/arr[@name='text']/str='This is a sentence just under <em>seventy</em> chars in length blah blah. '");
+    assertQ("default fragsize",
+        req("q", "text:seventy", "hl", "true", "hl.fragsize", "90"),
+        "//lst[@name='highlighting']/lst[@name='10']/arr[@name='text']/str='This is a sentence just under <em>seventy</em> chars in length blah blah. Next sentence is here.'");
   }
   
   public void testEncoder() {

solr/solr-ref-guide/src/highlighting.adoc

@@ -225,6 +225,14 @@ By default, the Unified Highlighter will usually pick the right offset source (s
 +
 The offset source can be explicitly configured to one of: `ANALYSIS`, `POSTINGS`, `POSTINGS_WITH_TERM_VECTORS`, or `TERM_VECTORS`.
 
+`hl.fragAlignRatio`::
+Fragment alignment can influence where the match in a passage is positioned. This floating point value is used to break the `hl.fragsize` around the match. The default value of `0.5` means to align the match to the middle. A value of `0.0` would mean to align the match to the left, while `1.0` to align it to the right. _Note: there are situations where the requested alignment is not plausible. This depends on the length of the match, the used breakiterator and the text content around the match._
+
+`hl.fragsizeIsMinimum`::
+By default this parameter is `false`. In this case the highlighter will accept a shorter fragment than `hl.fragsize`, if that is closer to the target than what the next longer choice would be.
++
+When treating `hl.fragsize` as a strict minimum length is acceptable, turning this on has some performance benefits to consider.
+
 `hl.tag.ellipsis`::
 By default, each snippet is returned as a separate value (as is done with the other highlighters). Set this parameter to instead return one string with this text as the delimiter. _Note: this is likely to be removed in the future._
 

solr/solrj/src/java/org/apache/solr/common/params/HighlightParams.java

@@ -47,6 +47,8 @@ public interface HighlightParams {
 
   // sizing
   public static final String FRAGSIZE    = HIGHLIGHT+".fragsize"; // OH, FVH, UH
+  public static final String FRAGSIZEISMINIMUM = HIGHLIGHT+".fragsizeIsMinimum"; // UH
+  public static final String FRAGALIGNRATIO = HIGHLIGHT+".fragAlignRatio"; // UH
   public static final String FRAGMENTER  = HIGHLIGHT+".fragmenter"; // OH
   public static final String INCREMENT   = HIGHLIGHT+".increment"; // OH
   public static final String REGEX       = "regex"; // OH