add assumption check

Author: gf2121

lucene/core/src/java/org/apache/lucene/codecs/lucene90/blocktree/Trie.java

@@ -142,7 +142,7 @@ long saveArcs(Node node, IndexOutput index, long startFP) throws IOException {
               | (output.hasTerms ? (1 << 5) : 0)
               | (output.floorData != null ? (1 << 6) : 0);
       index.writeByte(((byte) header));
-      writeLongNBytes(output.fp, outputFpBytes, index);
+      writeFpNBytes(output.fp, outputFpBytes, index);
       if (output.floorData != null) {
         index.writeBytes(output.floorData.bytes, output.floorData.offset, output.floorData.length);
       }
@@ -173,14 +173,14 @@ long saveArcs(Node node, IndexOutput index, long startFP) throws IOException {
       int header = SIGN_SINGLE_CHILDREN | (childFpBytes << 2) | (encodedOutputFpBytes << 5);
       index.writeByte((byte) header);
       index.writeByte((byte) node.children.getFirst().label);
-      writeLongNBytes(fpBuffer[0], childFpBytes, index);
+      writeFpNBytes(fpBuffer[0], childFpBytes, index);
       if (node.output != null) {
         Output output = node.output;
         long encodedFp =
             (output.floorData != null ? 0x01L : 0)
                 | (output.hasTerms ? 0x02L : 0)
                 | (output.fp << 2);
-        writeLongNBytes(encodedFp, encodedOutputFpBytes, index);
+        writeFpNBytes(encodedFp, encodedOutputFpBytes, index);
         if (output.floorData != null) {
           index.writeBytes(
               output.floorData.bytes, output.floorData.offset, output.floorData.length);
@@ -231,7 +231,7 @@ long saveArcs(Node node, IndexOutput index, long startFP) throws IOException {
       Output output = node.output;
       long encodedFp =
           (output.floorData != null ? 0x01L : 0) | (output.hasTerms ? 0x02L : 0) | (output.fp << 2);
-      writeLongNBytes(encodedFp, encodedOutputFpBytes, index);
+      writeFpNBytes(encodedFp, encodedOutputFpBytes, index);
     }
 
     long positionStartFp = index.getFilePointer();
@@ -244,7 +244,7 @@ long saveArcs(Node node, IndexOutput index, long startFP) throws IOException {
             + (index.getFilePointer() - positionStartFp);
 
     for (int i = 0; i < childrenNum; i++) {
-      writeLongNBytes(fpBuffer[i], childrenFpBytes, index);
+      writeFpNBytes(fpBuffer[i], childrenFpBytes, index);
     }
 
     if (node.output != null && node.output.floorData != null) {
@@ -259,7 +259,11 @@ private static int bytesRequired(long v) {
     return Math.max(1, Long.BYTES - (Long.numberOfLeadingZeros(v) >>> 3));
   }
 
-  private static void writeLongNBytes(long v, int n, DataOutput out) throws IOException {
+  private static void writeFpNBytes(long v, int n, DataOutput out) throws IOException {
+    if (n > 7) {
+      throw new IllegalArgumentException(
+          "term dictionary can not have file pointers bigger than 2^56, got: " + v);
+    }
     for (int i = 0; i < n; i++) {
       out.writeByte((byte) v);
       v >>= 8;

lucene/core/src/java/org/apache/lucene/codecs/lucene90/blocktree/TrieReader.java

@@ -83,13 +83,12 @@ private void load(Node node, long fp) throws IOException {
 
       // [n bytes] floor data
       // [n bytes] output fp
-      // [1bit] nothing | [1bit] has floor | [1bit] has terms | [3bit] output fp bytes  | [2bit]
+      // [1bit] nothing | [1bit] has floor | [1bit] has terms | [3bit] output fp bytes | [2bit]
       // sign
 
       node.childrenNum = 0;
       int fpBytes = (term >>> 2) & 0x07;
-      node.outputFp =
-          (termLong >>> 8) & bytesAsMask(fpBytes); // assumption fp for tip less than 56 bit
+      node.outputFp = (termLong >>> 8) & bytesAsMask(fpBytes);
       node.hasTerms = (term & 0x20) != 0;
       if ((term & 0x40) != 0) {
         node.floorDataFp = fp + 1 + fpBytes;
@@ -104,13 +103,13 @@ private void load(Node node, long fp) throws IOException {
 
       // [n bytes] floor data
       // [n bytes] encoded output fp | [n bytes] child fp | [1 byte] label
-      // [3bit] encoded output fp bytes | [3bit] child fp bytes | | [2bit] sign
+      // [3bit] encoded output fp bytes | [3bit] child fp bytes | [2bit] sign
 
       node.childrenNum = 1;
       int childFpBytes = (term >>> 2) & 0x07;
       int encodedOutputFpBytes = (term >>> 5) & 0x07;
-      node.childFp =
-          (termLong >>> 16) & bytesAsMask(childFpBytes); // assumption fp for tip less than 48 bit
+      long l = childFpBytes <= 6 ? termLong >>> 16 : access.readLong(fp + 2);
+      node.childFp = l & bytesAsMask(childFpBytes);
       node.minChildrenLabel = (term >>> 8) & 0xFF;
 
       if (encodedOutputFpBytes == 0) {

lucene/core/src/test/org/apache/lucene/codecs/lucene90/blocktree/TestTrie.java

@@ -42,9 +42,7 @@ public void testTrie() {
       BytesRef key = new BytesRef(randomBytes());
       Trie.Output value =
           new Trie.Output(
-              random().nextLong(Integer.MAX_VALUE),
-              random().nextBoolean(),
-              new BytesRef(randomBytes()));
+              random().nextLong(1L << 50), random().nextBoolean(), new BytesRef(randomBytes()));
       expected.put(key, value);
       trie.putAll(new Trie(key, value));
     }
@@ -66,7 +64,7 @@ public void testTrieLookup() throws IOException {
         BytesRef key = new BytesRef(randomBytes());
         Trie.Output value =
             new Trie.Output(
-                random().nextLong(Integer.MAX_VALUE),
+                random().nextLong(1L << 50),
                 random().nextBoolean(),
                 random().nextBoolean() ? null : new BytesRef(randomBytes()));
         expected.put(key, value);