Reimplement MultiMoreLikeThis using BigCompositeReader

src/main/java/it/cavallium/dbengine/lucene/ArrayIndexComparator.java

@@ -0,0 +1,25 @@
+package it.cavallium.dbengine.lucene;
+
+import java.util.Comparator;
+import org.apache.lucene.index.IndexReader;
+
+public class ArrayIndexComparator implements Comparator<IndexReader> {
+
+	private final Comparator<Object> comp;
+
+	public ArrayIndexComparator(IndexReader[] indexReaders) {
+		this.comp = Comparator.comparingInt(reader -> {
+			for (int i = 0; i < indexReaders.length; i++) {
+				if (indexReaders[i] == reader) {
+					return i;
+				}
+			}
+			throw new IllegalStateException();
+		});
+	}
+
+	@Override
+	public int compare(IndexReader o1, IndexReader o2) {
+		return comp.compare(o1, o2);
+	}
+}

src/main/java/it/cavallium/dbengine/lucene/LuceneUtils.java

@@ -20,6 +20,7 @@ import it.cavallium.dbengine.lucene.analyzer.NCharGramEdgeAnalyzer;
 import it.cavallium.dbengine.lucene.analyzer.TextFieldsAnalyzer;
 import it.cavallium.dbengine.lucene.analyzer.TextFieldsSimilarity;
 import it.cavallium.dbengine.lucene.analyzer.WordAnalyzer;
+import it.cavallium.dbengine.lucene.mlt.BigCompositeReader;
 import it.cavallium.dbengine.lucene.mlt.MultiMoreLikeThis;
 import it.cavallium.dbengine.lucene.searcher.ExponentialPageLimits;
 import it.cavallium.dbengine.lucene.searcher.LocalQueryParams;
@@ -50,6 +51,7 @@ import org.apache.lucene.index.IndexReader;
 import org.apache.lucene.index.IndexableField;
 import org.apache.lucene.search.BooleanClause.Occur;
 import org.apache.lucene.search.BooleanQuery;
+import org.apache.lucene.search.BooleanQuery.Builder;
 import org.apache.lucene.search.ConstantScoreQuery;
 import org.apache.lucene.search.FieldDoc;
 import org.apache.lucene.search.IndexSearcher;
@@ -557,13 +559,13 @@ public class LuceneUtils {
 					}
 					MultiMoreLikeThis mlt;
 					if (indexSearchers.size() == 1) {
-						mlt = new MultiMoreLikeThis(indexSearchers.get(0).getIndexReader(), null);
+						mlt = new MultiMoreLikeThis(new BigCompositeReader<>(indexSearchers.get(0).getIndexReader(), IndexReader[]::new), null);
 					} else {
 						IndexReader[] indexReaders = new IndexReader[indexSearchers.size()];
 						for (int i = 0, size = indexSearchers.size(); i < size; i++) {
 							indexReaders[i] = indexSearchers.get(i).getIndexReader();
 						}
-						mlt = new MultiMoreLikeThis(indexReaders, null);
+						mlt = new MultiMoreLikeThis(new BigCompositeReader<>(indexReaders, new ArrayIndexComparator(indexReaders)), null);
 					}
 					mlt.setAnalyzer(analyzer);
 					mlt.setFieldNames(mltDocumentFields.keySet().toArray(String[]::new));
@@ -583,7 +585,7 @@ public class LuceneUtils {
 					var mltQuery = mlt.like((Map) mltDocumentFields);
 					Query luceneQuery;
 					if (!(luceneAdditionalQuery instanceof MatchAllDocsQuery)) {
-						luceneQuery = new BooleanQuery.Builder()
+						luceneQuery = new Builder()
 								.add(mltQuery, Occur.MUST)
 								.add(new ConstantScoreQuery(luceneAdditionalQuery), Occur.MUST)
 								.build();

src/main/java/it/cavallium/dbengine/lucene/mlt/BigCompositeReader.java

@@ -0,0 +1,223 @@
+package it.cavallium.dbengine.lucene.mlt;
+
+import java.io.IOException;
+import java.math.BigInteger;
+import java.util.Arrays;
+import java.util.Collection;
+import java.util.Collections;
+import java.util.Comparator;
+import java.util.List;
+import java.util.Set;
+import java.util.concurrent.atomic.AtomicLong;
+import java.util.function.IntFunction;
+import java.util.function.ToIntFunction;
+import java.util.stream.Collectors;
+import java.util.stream.StreamSupport;
+import org.apache.lucene.document.Document;
+import org.apache.lucene.document.DocumentStoredFieldVisitor;
+import org.apache.lucene.index.Fields;
+import org.apache.lucene.index.IndexOptions;
+import org.apache.lucene.index.IndexReader;
+import org.apache.lucene.index.IndexReaderContext;
+import org.apache.lucene.index.LeafReaderContext;
+import org.apache.lucene.index.StoredFieldVisitor;
+import org.apache.lucene.index.Term;
+import org.apache.lucene.store.AlreadyClosedException;
+
+public class BigCompositeReader<R extends IndexReader> {
+
+	private static final long ACTUAL_MAX_DOCS = Long.MAX_VALUE - 10;
+	private final R[] subReaders;
+	protected final Comparator<R> subReadersSorter;
+	private final long[] starts;
+	private final long maxDoc;
+	private AtomicLong numDocs = new AtomicLong(-1);
+	private final List<R> subReadersList;
+
+	public BigCompositeReader(R subReader, IntFunction<R[]> arrayInstantiator) {
+		this(toArray(subReader, arrayInstantiator), null);
+	}
+
+	private static <R extends IndexReader> R[] toArray(R subReader, IntFunction<R[]> arrayInstantiator) {
+		var arr = arrayInstantiator.apply(1);
+		arr[0] = subReader;
+		return arr;
+	}
+
+	public BigCompositeReader(R[] subReaders, Comparator<R> subReadersSorter) {
+		if (subReadersSorter != null) {
+			Arrays.sort(subReaders, subReadersSorter);
+		}
+
+		this.subReaders = subReaders;
+		this.subReadersSorter = subReadersSorter;
+		this.subReadersList = Collections.unmodifiableList(Arrays.asList(subReaders));
+		this.starts = new long[subReaders.length + 1];
+		BigInteger maxDoc = BigInteger.ZERO;
+
+		for(int i = 0; i < subReaders.length; ++i) {
+			this.starts[i] = maxDoc.longValue();
+			IndexReader r = subReaders[i];
+			maxDoc = maxDoc.add(BigInteger.valueOf(r.maxDoc()));
+		}
+
+		if (maxDoc.compareTo(BigInteger.ZERO) < 0 || maxDoc.compareTo(BigInteger.valueOf(ACTUAL_MAX_DOCS)) > 0) {
+			throw new IllegalArgumentException("Too many documents: composite IndexReaders cannot exceed "
+					+ ACTUAL_MAX_DOCS + " but readers have total maxDoc=" + maxDoc);
+		} else {
+			this.maxDoc = maxDoc.longValueExact();
+			this.starts[subReaders.length] = this.maxDoc;
+		}
+	}
+
+	public static <T extends IndexReader> Collection<String> getIndexedFields(BigCompositeReader<T> readers) {
+		return readers.subReadersList
+				.stream()
+				.map(IndexReader::getContext)
+				.flatMap(l -> l.leaves().stream())
+				.flatMap((l) -> StreamSupport
+						.stream(l.reader().getFieldInfos().spliterator(), false)
+						.filter((fi) -> fi.getIndexOptions() != IndexOptions.NONE))
+				.map((fi) -> fi.name)
+				.collect(Collectors.toSet());
+	}
+
+	protected final List<? extends R> getSequentialSubReaders() {
+		return this.subReadersList;
+	}
+
+	private void ensureOpen() {
+		for (R subReader : subReaders) {
+			if (subReader.getRefCount() <= 0) {
+				throw new AlreadyClosedException("this IndexReader is closed");
+			}
+		}
+	}
+
+	public long getDocCount(String field) throws IOException {
+		this.ensureOpen();
+		long total = 0;
+		R[] var3 = this.subReaders;
+		long var4 = var3.length;
+
+		for(int var5 = 0; var5 < var4; ++var5) {
+			R reader = var3[var5];
+			int sub = reader.getDocCount(field);
+
+			assert sub >= 0;
+
+			assert sub <= reader.maxDoc();
+
+			total += sub;
+		}
+
+		return total;
+	}
+
+	public long docFreq(Term term) throws IOException {
+		this.ensureOpen();
+		long total = 0;
+
+		for(int i = 0; i < this.subReaders.length; ++i) {
+			int sub = this.subReaders[i].docFreq(term);
+
+			assert sub >= 0;
+
+			assert sub <= this.subReaders[i].getDocCount(term.field());
+
+			total += sub;
+		}
+
+		return total;
+	}
+
+	public long numDocs() {
+		long numDocs = this.numDocs.getOpaque();
+		if (numDocs == -1L) {
+			numDocs = 0L;
+			IndexReader[] var2 = this.subReaders;
+			int var3 = var2.length;
+
+			for(int var4 = 0; var4 < var3; ++var4) {
+				IndexReader r = var2[var4];
+				numDocs += r.numDocs();
+			}
+
+			assert numDocs >= 0L;
+
+			this.numDocs.set(numDocs);
+		}
+
+		return numDocs;
+	}
+
+	public Fields getTermVectors(long docID) throws IOException {
+		this.ensureOpen();
+		int i = this.readerIndex(docID);
+		return this.subReaders[i].getTermVectors(Math.toIntExact(docID - this.starts[i]));
+	}
+
+	protected final int readerIndex(long docID) {
+		if (docID >= 0 && docID < this.maxDoc) {
+			return subIndex(docID, this.starts);
+		} else {
+			throw new IllegalArgumentException("docID must be >= 0 and < maxDoc=" + this.maxDoc + " (got docID=" + docID + ")");
+		}
+	}
+
+	public static int subIndex(long n, long[] docStarts) {
+		int size = docStarts.length;
+		int lo = 0;
+		int hi = size - 1;
+
+		while(hi >= lo) {
+			int mid = lo + hi >>> 1;
+			long midValue = docStarts[mid];
+			if (n < midValue) {
+				hi = mid - 1;
+			} else {
+				if (n <= midValue) {
+					while(mid + 1 < size && docStarts[mid + 1] == midValue) {
+						++mid;
+					}
+
+					return mid;
+				}
+
+				lo = mid + 1;
+			}
+		}
+
+		return hi;
+	}
+
+	protected final long readerBase(int readerIndex) {
+		if (readerIndex >= 0L && readerIndex < this.subReaders.length) {
+			return this.starts[readerIndex];
+		} else {
+			throw new IllegalArgumentException("readerIndex must be >= 0 and < getSequentialSubReaders().size()");
+		}
+	}
+
+	public final void document(long docID, StoredFieldVisitor visitor) throws IOException {
+		this.ensureOpen();
+		int i = this.readerIndex(docID);
+		this.subReaders[i].document(Math.toIntExact(docID - this.starts[i]), visitor);
+	}
+
+	public final Document document(long docID) throws IOException {
+		DocumentStoredFieldVisitor visitor = new DocumentStoredFieldVisitor();
+		this.document(docID, (StoredFieldVisitor)visitor);
+		return visitor.getDocument();
+	}
+
+	public final Document document(long docID, Set<String> fieldsToLoad) throws IOException {
+		DocumentStoredFieldVisitor visitor = new DocumentStoredFieldVisitor(fieldsToLoad);
+		this.document(docID, (StoredFieldVisitor)visitor);
+		return visitor.getDocument();
+	}
+
+	public long maxDoc() {
+		return this.maxDoc;
+	}
+}

src/main/java/it/cavallium/dbengine/lucene/mlt/MultiMoreLikeThis.java

@@ -29,9 +29,7 @@ import org.apache.lucene.analysis.TokenStream;
 import org.apache.lucene.analysis.tokenattributes.CharTermAttribute;
 import org.apache.lucene.analysis.tokenattributes.TermFrequencyAttribute;
 import org.apache.lucene.document.Document;
-import org.apache.lucene.index.FieldInfos;
 import org.apache.lucene.index.Fields;
-import org.apache.lucene.index.IndexReader;
 import org.apache.lucene.index.IndexableField;
 import org.apache.lucene.index.Term;
 import org.apache.lucene.index.Terms;
@@ -52,7 +50,7 @@ import org.apache.lucene.util.PriorityQueue;
  * Generate "more like this" similarity queries. Based on this mail:
  *
  * <pre><code>
- * Lucene does let you access the document frequency of terms, with IndexReader.docFreq().
+ * Lucene does let you access the document frequency of terms, with BigCompositeReader.docFreq().
  * Term frequencies can be computed by re-tokenizing the text, which, for a single document,
  * is usually fast enough.  But looking up the docFreq() of every term in the document is
  * probably too slow.
@@ -86,7 +84,7 @@ import org.apache.lucene.util.PriorityQueue;
  * usage is as follows. The bold fragment is specific to this class. <br>
  *
  * <pre class="prettyprint">
- * IndexReader ir = ...
+ * BigCompositeReader ir = ...
  * IndexSearcher is = ...
  *
  * MoreLikeThis mlt = new MoreLikeThis(ir);
@@ -137,7 +135,7 @@ import org.apache.lucene.util.PriorityQueue;
  * <pre>
  * Changes: Mark Harwood 29/02/04
  * Some bugfixing, some refactoring, some optimisation.
- * - bugfix: retrieveTerms(int docNum) was not working for indexes without a termvector -added missing code
+ * - bugfix: retrieveTerms(long docNum) was not working for indexes without a termvector -added missing code
  * - bugfix: No significant terms being created for fields with a termvector - because
  * was only counting one occurrence per term/field pair in calculations(ie not including frequency info from TermVector)
  * - refactor: moved common code into isNoiseWord()
@@ -153,7 +151,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @see #getMaxNumTokensParsed
 	 */
-	public static final int DEFAULT_MAX_NUM_TOKENS_PARSED = 5000;
+	public static final long DEFAULT_MAX_NUM_TOKENS_PARSED = 5000;
 
 	/**
 	 * Ignore terms with less than this frequency in the source doc.
@@ -161,7 +159,7 @@ public final class MultiMoreLikeThis {
 	 * @see #getMinTermFreq
 	 * @see #setMinTermFreq
 	 */
-	public static final int DEFAULT_MIN_TERM_FREQ = 2;
+	public static final long DEFAULT_MIN_TERM_FREQ = 2;
 
 	/**
 	 * Ignore words which do not occur in at least this many docs.
@@ -178,7 +176,7 @@ public final class MultiMoreLikeThis {
 	 * @see #setMaxDocFreq
 	 * @see #setMaxDocFreqPct
 	 */
-	public static final long DEFAULT_MAX_DOC_FREQ = Long.MAX_VALUE;
+	public static final long DEFAULT_MAX_DOC_FREQ = java.lang.Long.MAX_VALUE;
 
 	/**
 	 * Boost terms in query based on score.
@@ -200,7 +198,7 @@ public final class MultiMoreLikeThis {
 	 * @see #getMinWordLen
 	 * @see #setMinWordLen
 	 */
-	public static final int DEFAULT_MIN_WORD_LENGTH = 0;
+	public static final long DEFAULT_MIN_WORD_LENGTH = 0;
 
 	/**
 	 * Ignore words greater than this length or if 0 then this has no effect.
@@ -208,7 +206,7 @@ public final class MultiMoreLikeThis {
 	 * @see #getMaxWordLen
 	 * @see #setMaxWordLen
 	 */
-	public static final int DEFAULT_MAX_WORD_LENGTH = 0;
+	public static final long DEFAULT_MAX_WORD_LENGTH = 0;
 
 	/**
 	 * Default set of stopwords. If null means to allow stop words.
@@ -228,13 +226,13 @@ public final class MultiMoreLikeThis {
 	 * @see #getMaxQueryTerms
 	 * @see #setMaxQueryTerms
 	 */
-	public static final int DEFAULT_MAX_QUERY_TERMS = 25;
+	public static final long DEFAULT_MAX_QUERY_TERMS = 25;
 
 	/** Analyzer that will be used to parse the doc. */
 	private Analyzer analyzer = null;
 
 	/** Ignore words less frequent that this. */
-	private int minTermFreq = DEFAULT_MIN_TERM_FREQ;
+	private long minTermFreq = DEFAULT_MIN_TERM_FREQ;
 
 	/** Ignore words which do not occur in at least this many docs. */
 	private long minDocFreq = DEFAULT_MIN_DOC_FREQ;
@@ -252,27 +250,22 @@ public final class MultiMoreLikeThis {
 	 * The maximum number of tokens to parse in each example doc field that is not stored with
 	 * TermVector support
 	 */
-	private int maxNumTokensParsed = DEFAULT_MAX_NUM_TOKENS_PARSED;
+	private long maxNumTokensParsed = DEFAULT_MAX_NUM_TOKENS_PARSED;
 
 	/** Ignore words if less than this len. */
-	private int minWordLen = DEFAULT_MIN_WORD_LENGTH;
+	private long minWordLen = DEFAULT_MIN_WORD_LENGTH;
 
 	/** Ignore words if greater than this len. */
-	private int maxWordLen = DEFAULT_MAX_WORD_LENGTH;
+	private long maxWordLen = DEFAULT_MAX_WORD_LENGTH;
 
 	/** Don't return a query longer than this. */
-	private int maxQueryTerms = DEFAULT_MAX_QUERY_TERMS;
+	private long maxQueryTerms = DEFAULT_MAX_QUERY_TERMS;
 
 	/** For idf() calculations. */
 	private TFIDFSimilarity similarity; // = new DefaultSimilarity();
 
-	/** IndexReader to use */
+	/** BigCompositeReader to use */
-	private final IndexReader ir;
+	private final BigCompositeReader<?> ir;
-
-	/**
-	 * IndexReader array to use when multi-searchers are used.
-	 */
-	private final IndexReader[] irArray;
 
 	/** Boost factor to use when boosting the terms */
 	private float boostFactor = 1;
@@ -296,28 +289,12 @@ public final class MultiMoreLikeThis {
 		this.boostFactor = boostFactor;
 	}
 
-	/** Constructor requiring an IndexReader. */
+	/** Constructor requiring a BigCompositeReader. */
-	public MultiMoreLikeThis(IndexReader ir) {
+	public MultiMoreLikeThis(BigCompositeReader<?> ir) {
 		this(ir, new ClassicSimilarity());
 	}
 
-	public MultiMoreLikeThis(IndexReader ir, TFIDFSimilarity sim) {
+	public MultiMoreLikeThis(BigCompositeReader<?> ir, TFIDFSimilarity sim) {
-		this(ir, null, sim);
-	}
-
-	public MultiMoreLikeThis(IndexReader[] irArray) {
-		this(irArray, new ClassicSimilarity());
-	}
-
-	public MultiMoreLikeThis(IndexReader[] irArray, TFIDFSimilarity sim) {
-		this(null, irArray, sim);
-	}
-
-	private MultiMoreLikeThis(IndexReader ir, IndexReader[] irArray, TFIDFSimilarity sim) {
-		if ((ir == null) == (irArray == null)) {
-			throw new IllegalArgumentException();
-		}
-		this.irArray = irArray;
 		this.ir = ir;
 		this.similarity = sim;
 	}
@@ -342,7 +319,7 @@ public final class MultiMoreLikeThis {
 
 	/**
 	 * Sets the analyzer to use. An analyzer is not required for generating a query with the {@link
-	 * #like(int)} method, all other 'like' methods require an analyzer.
+	 * #like(long)} method, all other 'like' methods require an analyzer.
 	 *
 	 * @param analyzer the analyzer to use to tokenize text.
 	 */
@@ -356,7 +333,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @return the frequency below which terms will be ignored in the source doc.
 	 */
-	public int getMinTermFreq() {
+	public long getMinTermFreq() {
 		return minTermFreq;
 	}
 
@@ -365,7 +342,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @param minTermFreq the frequency below which terms will be ignored in the source doc.
 	 */
-	public void setMinTermFreq(int minTermFreq) {
+	public void setMinTermFreq(long minTermFreq) {
 		this.minTermFreq = minTermFreq;
 	}
 
@@ -423,17 +400,8 @@ public final class MultiMoreLikeThis {
 	 * @param maxPercentage the maximum percentage of documents (0-100) that a term may appear in to
 	 *     be still considered relevant.
 	 */
-	public void setMaxDocFreqPct(int maxPercentage) {
+	public void setMaxDocFreqPct(long maxPercentage) {
-		long maxDoc;
+		setMaxDocFreq((maxPercentage) * ir.maxDoc() / 100L);
-		if (irArray == null) {
-			maxDoc = ir.maxDoc();
-		} else {
-			maxDoc = 0L;
-			for (IndexReader ir : irArray) {
-				maxDoc += ir.maxDoc();
-			}
-		}
-		setMaxDocFreq(Math.toIntExact((long) maxPercentage * maxDoc / 100L));
 	}
 
 	/**
@@ -469,7 +437,7 @@ public final class MultiMoreLikeThis {
 
 	/**
 	 * Sets the field names that will be used when generating the 'More Like This' query. Set this to
-	 * null for the field names to be determined at runtime from the IndexReader provided in the
+	 * null for the field names to be determined at runtime from the BigCompositeReader provided in the
 	 * constructor.
 	 *
 	 * @param fieldNames the field names that will be used when generating the 'More Like This' query.
@@ -484,7 +452,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @return the minimum word length below which words will be ignored.
 	 */
-	public int getMinWordLen() {
+	public long getMinWordLen() {
 		return minWordLen;
 	}
 
@@ -493,7 +461,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @param minWordLen the minimum word length below which words will be ignored.
 	 */
-	public void setMinWordLen(int minWordLen) {
+	public void setMinWordLen(long minWordLen) {
 		this.minWordLen = minWordLen;
 	}
 
@@ -503,7 +471,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @return the maximum word length above which words will be ignored.
 	 */
-	public int getMaxWordLen() {
+	public long getMaxWordLen() {
 		return maxWordLen;
 	}
 
@@ -512,7 +480,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @param maxWordLen the maximum word length above which words will be ignored.
 	 */
-	public void setMaxWordLen(int maxWordLen) {
+	public void setMaxWordLen(long maxWordLen) {
 		this.maxWordLen = maxWordLen;
 	}
 
@@ -544,7 +512,7 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @return the maximum number of query terms that will be included in any generated query.
 	 */
-	public int getMaxQueryTerms() {
+	public long getMaxQueryTerms() {
 		return maxQueryTerms;
 	}
 
@@ -554,7 +522,7 @@ public final class MultiMoreLikeThis {
 	 * @param maxQueryTerms the maximum number of query terms that will be included in any generated
 	 *     query.
 	 */
-	public void setMaxQueryTerms(int maxQueryTerms) {
+	public void setMaxQueryTerms(long maxQueryTerms) {
 		this.maxQueryTerms = maxQueryTerms;
 	}
 
@@ -563,7 +531,7 @@ public final class MultiMoreLikeThis {
 	 *     TermVector support
 	 * @see #DEFAULT_MAX_NUM_TOKENS_PARSED
 	 */
-	public int getMaxNumTokensParsed() {
+	public long getMaxNumTokensParsed() {
 		return maxNumTokensParsed;
 	}
 
@@ -571,7 +539,7 @@ public final class MultiMoreLikeThis {
 	 * @param i The maximum number of tokens to parse in each example doc field that is not stored
 	 *     with TermVector support
 	 */
-	public void setMaxNumTokensParsed(int i) {
+	public void setMaxNumTokensParsed(long i) {
 		maxNumTokensParsed = i;
 	}
 
@@ -581,18 +549,11 @@ public final class MultiMoreLikeThis {
 	 * @param docNum the documentID of the lucene doc to generate the 'More Like This" query for.
 	 * @return a query that will return docs like the passed lucene document ID.
 	 */
-	public Query like(int docNum) throws IOException {
+	public Query like(long docNum) throws IOException {
 		if (fieldNames == null) {
 			// gather list of valid fields from lucene
 			Collection<String> fields;
-			if (irArray == null) {
+			fields = BigCompositeReader.getIndexedFields(ir);
-				fields = FieldInfos.getIndexedFields(ir);
-			} else {
-				fields = new ArrayList<>();
-				for (IndexReader ir : irArray) {
-					fields.addAll(FieldInfos.getIndexedFields(ir));
-				}
-			}
 			fieldNames = fields.toArray(String[]::new);
 		}
 
@@ -606,15 +567,7 @@ public final class MultiMoreLikeThis {
 	public Query like(Map<String, Collection<Object>> filteredDocument) throws IOException {
 		if (fieldNames == null) {
 			// gather list of valid fields from lucene
-			Collection<String> fields;
+			Collection<String> fields = BigCompositeReader.getIndexedFields(ir);
-			if (irArray == null) {
-				fields = FieldInfos.getIndexedFields(ir);
-			} else {
-				fields = new ArrayList<>();
-				for (IndexReader ir : irArray) {
-					fields.addAll(FieldInfos.getIndexedFields(ir));
-				}
-			}
 			fieldNames = fields.toArray(String[]::new);
 		}
 		return createQuery(retrieveTerms(filteredDocument));
@@ -627,7 +580,7 @@ public final class MultiMoreLikeThis {
 	 * @return a query that will return docs like the passed Readers.
 	 */
 	public Query like(String fieldName, Reader... readers) throws IOException {
-		Map<String, Map<String, Int>> perFieldTermFrequencies = new HashMap<>();
+		Map<String, Map<String, Long>> perFieldTermFrequencies = new HashMap<>();
 		for (Reader r : readers) {
 			addTermFrequencies(r, perFieldTermFrequencies, fieldName);
 		}
@@ -669,48 +622,28 @@ public final class MultiMoreLikeThis {
 	 *     objects as the values.
 	 */
 	private PriorityQueue<ScoreTerm> createQueue(
-			Map<String, Map<String, Int>> perFieldTermFrequencies) throws IOException {
+			Map<String, Map<String, Long>> perFieldTermFrequencies) throws IOException {
 		// have collected all words in doc and their freqs
-		final int limit = Math.min(maxQueryTerms, this.getTermsCount(perFieldTermFrequencies));
+		final long limit = Math.min(maxQueryTerms, this.getTermsCount(perFieldTermFrequencies));
-		FreqQ queue = new FreqQ(limit); // will order words by score
+		FreqQ queue = new FreqQ(Math.toIntExact(limit)); // will order words by score
-		for (Map.Entry<String, Map<String, Int>> entry : perFieldTermFrequencies.entrySet()) {
+		for (Map.Entry<String, Map<String, Long>> entry : perFieldTermFrequencies.entrySet()) {
-			Map<String, Int> perWordTermFrequencies = entry.getValue();
+			Map<String, Long> perWordTermFrequencies = entry.getValue();
 			String fieldName = entry.getKey();
 
-			long numDocs;
+			long numDocs = ir.getDocCount(fieldName);
-			if (irArray == null) {
+			if (numDocs == -1) {
-				numDocs = ir.getDocCount(fieldName);
+				numDocs = ir.numDocs();
-				if (numDocs == -1) {
-					numDocs = ir.numDocs();
-				}
-			} else {
-				numDocs = 0L;
-				for (IndexReader ir : irArray) {
-					long localNumDocs = ir.getDocCount(fieldName);
-					if (localNumDocs == -1) {
-						localNumDocs = ir.numDocs();
-					}
-					numDocs += localNumDocs;
-				}
 			}
 
-			for (Map.Entry<String, Int> tfEntry : perWordTermFrequencies.entrySet()) { // for every word
+			for (Map.Entry<String, Long> tfEntry : perWordTermFrequencies.entrySet()) { // for every word
 				String word = tfEntry.getKey();
-				int tf = tfEntry.getValue().x; // term freq in the source doc
+				long tf = tfEntry.getValue().x; // term freq in the source doc
 				if (minTermFreq > 0 && tf < minTermFreq) {
 					continue; // filter out words that don't occur enough times in the source
 				}
 
-				long docFreq;
 				var fieldTerm = new Term(fieldName, word);
-				if (irArray == null) {
+				long docFreq = ir.docFreq(fieldTerm);
-					docFreq = ir.docFreq(fieldTerm);
-				} else {
-					docFreq = 0;
-					for (IndexReader ir : irArray) {
-						docFreq += ir.docFreq(fieldTerm);
-					}
-				}
 
 				if (minDocFreq > 0L && docFreq < minDocFreq) {
 					continue; // filter out words that don't occur in enough docs
@@ -743,10 +676,10 @@ public final class MultiMoreLikeThis {
 		return queue;
 	}
 
-	private int getTermsCount(Map<String, Map<String, Int>> perFieldTermFrequencies) {
+	private long getTermsCount(Map<String, Map<String, Long>> perFieldTermFrequencies) {
-		int totalTermsCount = 0;
+		long totalTermsCount = 0;
-		Collection<Map<String, Int>> values = perFieldTermFrequencies.values();
+		Collection<Map<String, Long>> values = perFieldTermFrequencies.values();
-		for (Map<String, Int> perWordTermFrequencies : values) {
+		for (Map<String, Long> perWordTermFrequencies : values) {
 			totalTermsCount += perWordTermFrequencies.size();
 		}
 		return totalTermsCount;
@@ -776,20 +709,14 @@ public final class MultiMoreLikeThis {
 	 *
 	 * @param docNum the id of the lucene document from which to find terms
 	 */
-	private PriorityQueue<ScoreTerm> retrieveTerms(int docNum) throws IOException {
+	private PriorityQueue<ScoreTerm> retrieveTerms(long docNum) throws IOException {
-		Map<String, Map<String, Int>> field2termFreqMap = new HashMap<>();
+		Map<String, Map<String, Long>> field2termFreqMap = new HashMap<>();
-		if (irArray == null) {
+		retrieveTermsOfIndexReader(ir, docNum, field2termFreqMap);
-			retrieveTermsOfIndexReader(ir, docNum, field2termFreqMap);
-		} else {
-			for (IndexReader ir : irArray) {
-				retrieveTermsOfIndexReader(ir, docNum, field2termFreqMap);
-			}
-		}
 
 		return createQueue(field2termFreqMap);
 	}
 
-	private void retrieveTermsOfIndexReader(IndexReader ir, int docNum, Map<String, Map<String, Int>> field2termFreqMap)
+	private void retrieveTermsOfIndexReader(BigCompositeReader<?> ir, long docNum, Map<String, Map<String, Long>> field2termFreqMap)
 			throws IOException {
 		for (String fieldName : fieldNames) {
 			final Fields vectors = ir.getTermVectors(docNum);
@@ -818,7 +745,7 @@ public final class MultiMoreLikeThis {
 
 	private PriorityQueue<ScoreTerm> retrieveTerms(Map<String, Collection<Object>> field2fieldValues)
 			throws IOException {
-		Map<String, Map<String, Int>> field2termFreqMap = new HashMap<>();
+		Map<String, Map<String, Long>> field2termFreqMap = new HashMap<>();
 		for (String fieldName : fieldNames) {
 			Collection<Object> fieldValues = field2fieldValues.get(fieldName);
 			if (fieldValues == null) {
@@ -840,9 +767,9 @@ public final class MultiMoreLikeThis {
 	 * @param vector List of terms and their frequencies for a doc/field
 	 */
 	private void addTermFrequencies(
-			Map<String, Map<String, Int>> field2termFreqMap, Terms vector, String fieldName)
+			Map<String, Map<String, Long>> field2termFreqMap, Terms vector, String fieldName)
 			throws IOException {
-		Map<String, Int> termFreqMap =
+		Map<String, Long> termFreqMap =
 				field2termFreqMap.computeIfAbsent(fieldName, k -> new HashMap<>());
 		final TermsEnum termsEnum = vector.iterator();
 		final CharsRefBuilder spare = new CharsRefBuilder();
@@ -853,12 +780,12 @@ public final class MultiMoreLikeThis {
 			if (isNoiseWord(term)) {
 				continue;
 			}
-			final int freq = (int) termsEnum.totalTermFreq();
+			final long freq = termsEnum.totalTermFreq();
 
 			// increment frequency
-			Int cnt = termFreqMap.get(term);
+			Long cnt = termFreqMap.get(term);
 			if (cnt == null) {
-				cnt = new Int();
+				cnt = new Long();
 				termFreqMap.put(term, cnt);
 				cnt.x = freq;
 			} else {
@@ -875,16 +802,16 @@ public final class MultiMoreLikeThis {
 	 * @param fieldName Used by analyzer for any special per-field analysis
 	 */
 	private void addTermFrequencies(
-			Reader r, Map<String, Map<String, Int>> perFieldTermFrequencies, String fieldName)
+			Reader r, Map<String, Map<String, Long>> perFieldTermFrequencies, String fieldName)
 			throws IOException {
 		if (analyzer == null) {
 			throw new UnsupportedOperationException(
 					"To use MoreLikeThis without " + "term vectors, you must provide an Analyzer");
 		}
-		Map<String, Int> termFreqMap =
+		Map<String, Long> termFreqMap =
 				perFieldTermFrequencies.computeIfAbsent(fieldName, k -> new HashMap<>());
 		try (TokenStream ts = analyzer.tokenStream(fieldName, r)) {
-			int tokenCount = 0;
+			long tokenCount = 0;
 			// for every token
 			CharTermAttribute termAtt = ts.addAttribute(CharTermAttribute.class);
 			TermFrequencyAttribute tfAtt = ts.addAttribute(TermFrequencyAttribute.class);
@@ -900,9 +827,9 @@ public final class MultiMoreLikeThis {
 				}
 
 				// increment frequency
-				Int cnt = termFreqMap.get(word);
+				Long cnt = termFreqMap.get(word);
 				if (cnt == null) {
-					termFreqMap.put(word, new Int(tfAtt.getTermFrequency()));
+					termFreqMap.put(word, new Long(tfAtt.getTermFrequency()));
 				} else {
 					cnt.x += tfAtt.getTermFrequency();
 				}
@@ -918,7 +845,7 @@ public final class MultiMoreLikeThis {
 	 * @return true if should be ignored, false if should be used in further analysis
 	 */
 	private boolean isNoiseWord(String term) {
-		int len = term.length();
+		long len = term.length();
 		if (minWordLen > 0 && len < minWordLen) {
 			return true;
 		}
@@ -953,19 +880,19 @@ public final class MultiMoreLikeThis {
 	 * @see #retrieveInterestingTerms
 	 */
 	private PriorityQueue<ScoreTerm> retrieveTerms(Reader r, String fieldName) throws IOException {
-		Map<String, Map<String, Int>> field2termFreqMap = new HashMap<>();
+		Map<String, Map<String, Long>> field2termFreqMap = new HashMap<>();
 		addTermFrequencies(r, field2termFreqMap, fieldName);
 		return createQueue(field2termFreqMap);
 	}
 
 	/** @see #retrieveInterestingTerms(java.io.Reader, String) */
-	public String[] retrieveInterestingTerms(int docNum) throws IOException {
+	public String[] retrieveInterestingTerms(long docNum) throws IOException {
-		ArrayList<String> al = new ArrayList<>(maxQueryTerms);
+		ArrayList<String> al = new ArrayList<>(Math.toIntExact(maxQueryTerms));
 		PriorityQueue<ScoreTerm> pq = retrieveTerms(docNum);
 		ScoreTerm scoreTerm;
 		// have to be careful, retrieveTerms returns all words but that's probably not useful to our
 		// caller...
-		int lim = maxQueryTerms;
+		long lim = maxQueryTerms;
 		// we just want to return the top words
 		while (((scoreTerm = pq.pop()) != null) && lim-- > 0) {
 			al.add(scoreTerm.word); // the 1st entry is the interesting word
@@ -985,12 +912,12 @@ public final class MultiMoreLikeThis {
 	 * @see #setMaxQueryTerms
 	 */
 	public String[] retrieveInterestingTerms(Reader r, String fieldName) throws IOException {
-		ArrayList<String> al = new ArrayList<>(maxQueryTerms);
+		ArrayList<String> al = new ArrayList<>(Math.toIntExact(maxQueryTerms));
 		PriorityQueue<ScoreTerm> pq = retrieveTerms(r, fieldName);
 		ScoreTerm scoreTerm;
 		// have to be careful, retrieveTerms returns all words but that's probably not useful to our
 		// caller...
-		int lim = maxQueryTerms;
+		long lim = maxQueryTerms;
 		// we just want to return the top words
 		while (((scoreTerm = pq.pop()) != null) && lim-- > 0) {
 			al.add(scoreTerm.word); // the 1st entry is the interesting word
@@ -1031,14 +958,14 @@ public final class MultiMoreLikeThis {
 	}
 
 	/** Use for frequencies and to avoid renewing Integers. */
-	private static class Int {
+	private static class Long {
-		int x;
+		long x;
 
-		Int() {
+		Long() {
-			this(1);
+			this(1L);
 		}
 
-		Int(int initialValue) {
+		Long(long initialValue) {
 			x = initialValue;
 		}
 	}