Fix returned Impacts when frequencies are not indexed (#15263)

This updates postings to always return impacts with freq=1 and norm=1 when frequencies are not indexed.

Author: iverase

lucene/CHANGES.txt

@@ -7,7 +7,9 @@ http://s.apache.org/luceneversions
 
 Bug Fixes
 ---------------------
-(No changes)
+* GITHUB#15263: Fix the returned Impact returned from Lucene103PostingsReader when frequencies
+  are not indexed. It was returning a wrong frequency in that case affecting scoring which
+  might led to performance issues. (Ignacio Vera)
 
 ======================= Lucene 10.3.0 =======================
 

lucene/core/src/java/org/apache/lucene/codecs/lucene103/Lucene103PostingsReader.java

@@ -74,6 +74,8 @@ public final class Lucene103PostingsReader extends PostingsReaderBase {
   // less than 128 docs left to evaluate anyway.
   private static final List<Impact> DUMMY_IMPACTS =
       Collections.singletonList(new Impact(Integer.MAX_VALUE, 1L));
+  // impacts when there is no frequency, max frequency is 1.
+  private static final List<Impact> IMPACTS_NO_FREQ = Collections.singletonList(new Impact(1, 1L));
 
   private final IndexInput docIn;
   private final IndexInput posIn;
@@ -1381,8 +1383,9 @@ public List<Impact> getImpacts(int level) {
               if (level == 1) {
                 return readImpacts(level1SerializedImpacts, level1Impacts);
               }
+              return DUMMY_IMPACTS;
             }
-            return DUMMY_IMPACTS;
+            return IMPACTS_NO_FREQ;
           }
 
           private List<Impact> readImpacts(BytesRef serialized, MutableImpactList impactsList) {

lucene/core/src/test/org/apache/lucene/codecs/lucene103/TestLucene103PostingsFormat.java

@@ -29,8 +29,12 @@
 import org.apache.lucene.document.Field;
 import org.apache.lucene.index.DirectoryReader;
 import org.apache.lucene.index.Impact;
+import org.apache.lucene.index.ImpactsEnum;
 import org.apache.lucene.index.IndexWriter;
 import org.apache.lucene.index.IndexWriterConfig;
+import org.apache.lucene.index.LeafReader;
+import org.apache.lucene.index.PostingsEnum;
+import org.apache.lucene.index.TermsEnum;
 import org.apache.lucene.store.ByteArrayDataInput;
 import org.apache.lucene.store.ByteArrayDataOutput;
 import org.apache.lucene.store.Directory;
@@ -39,7 +43,9 @@
 import org.apache.lucene.store.IndexOutput;
 import org.apache.lucene.tests.analysis.MockAnalyzer;
 import org.apache.lucene.tests.index.BasePostingsFormatTestCase;
+import org.apache.lucene.tests.index.RandomIndexWriter;
 import org.apache.lucene.tests.util.TestUtil;
+import org.apache.lucene.util.BytesRef;
 
 public class TestLucene103PostingsFormat extends BasePostingsFormatTestCase {
 
@@ -154,4 +160,26 @@ private void doTestImpactSerialization(List<Impact> impacts) throws IOException
       }
     }
   }
+
+  public void testImpactsNoFreqs() throws Exception {
+    try (Directory dir = newDirectory()) {
+      IndexWriterConfig iwc = newIndexWriterConfig(new MockAnalyzer(random()));
+      iwc.setCodec(getCodec());
+      try (RandomIndexWriter iw = new RandomIndexWriter(random(), dir, iwc)) {
+        Document doc = new Document();
+        doc.add(newStringField("field", "value", Field.Store.NO));
+        iw.addDocument(doc);
+        try (DirectoryReader ir = iw.getReader()) {
+          LeafReader ar = getOnlyLeafReader(ir);
+          TermsEnum termsEnum = ar.terms("field").iterator();
+          termsEnum.seekExact(new BytesRef("value"));
+          ImpactsEnum impactsEnum = termsEnum.impacts(PostingsEnum.FREQS);
+          List<Impact> impacts = impactsEnum.getImpacts().getImpacts(0);
+          assertEquals(1, impacts.size());
+          assertEquals(1, impacts.get(0).freq);
+          assertEquals(1L, impacts.get(0).norm);
+        }
+      }
+    }
+  }
 }