core: improve error handling in decoder & transcoder

core/src/main/java/eu/ostrzyciel/jelly/core/internal/NameDecoderImpl.java

@@ -62,7 +62,7 @@ public NameDecoderImpl(int prefixTableSize, int nameTableSize, Function<String,
     /**
      * Update the name table with a new entry.
      * @param nameEntry name row
-     * @throws ArrayIndexOutOfBoundsException if the identifier is out of bounds
+     * @throws RdfProtoDeserializationError if the identifier is out of bounds
      */
     @Override
     public void updateNames(RdfNameEntry nameEntry) {
@@ -72,33 +72,44 @@ public void updateNames(RdfNameEntry nameEntry) {
         //   else lastNameIdSet = id;
         // Same code is used in the methods below.
         lastNameIdSet = ((lastNameIdSet + 1) & ((id - 1) >> 31)) + id;
-        NameLookupEntry entry = nameLookup[lastNameIdSet];
-        entry.name = nameEntry.value();
-        // Enough to invalidate the last IRI – we don't have to touch the serial number.
-        entry.lastPrefixId = 0;
-        // Set to null is required to avoid a false positive in the decode method for cases without a prefix.
-        entry.lastIri = null;
+        try {
+            NameLookupEntry entry = nameLookup[lastNameIdSet];
+            entry.name = nameEntry.value();
+            // Enough to invalidate the last IRI – we don't have to touch the serial number.
+            entry.lastPrefixId = 0;
+            // Set to null is required to avoid a false positive in the decode method for cases without a prefix.
+            entry.lastIri = null;
+        } catch (ArrayIndexOutOfBoundsException | NullPointerException e) {
+            throw JellyExceptions.rdfProtoDeserializationError(
+                "Name entry with ID " + id + " is out of bounds of the name lookup table."
+            );
+        }
     }
 
     /**
      * Update the prefix table with a new entry.
      * @param prefixEntry prefix row
-     * @throws ArrayIndexOutOfBoundsException if the identifier is out of bounds
+     * @throws RdfProtoDeserializationError if the identifier is out of bounds
      */
     @Override
     public void updatePrefixes(RdfPrefixEntry prefixEntry) {
         int id = prefixEntry.id();
         lastPrefixIdSet = ((lastPrefixIdSet + 1) & ((id - 1) >> 31)) + id;
-        PrefixLookupEntry entry = prefixLookup[lastPrefixIdSet];
-        entry.prefix = prefixEntry.value();
-        entry.serial++;
+        try {
+            PrefixLookupEntry entry = prefixLookup[lastPrefixIdSet];
+            entry.prefix = prefixEntry.value();
+            entry.serial++;
+        } catch (ArrayIndexOutOfBoundsException | NullPointerException e) {
+            throw JellyExceptions.rdfProtoDeserializationError(
+                "Prefix entry with ID " + id + " is out of bounds of the prefix lookup table."
+            );
+        }
     }
 
     /**
      * Reconstruct an IRI from its prefix and name ids.
      * @param iri IRI row from the Jelly proto
      * @return full IRI combining the prefix and the name
-     * @throws ArrayIndexOutOfBoundsException if IRI had indices out of lookup table bounds
      * @throws RdfProtoDeserializationError if the IRI reference is invalid
      * @throws NullPointerException if the IRI reference is invalid
      */
@@ -107,8 +118,18 @@ public void updatePrefixes(RdfPrefixEntry prefixEntry) {
     public TIri decode(RdfIri iri) {
         int nameId = iri.nameId();
         lastNameIdReference = ((lastNameIdReference + 1) & ((nameId - 1) >> 31)) + nameId;
-        NameLookupEntry nameEntry = nameLookup[lastNameIdReference];
-
+        NameLookupEntry nameEntry;
+        try {
+            nameEntry = nameLookup[lastNameIdReference];
+        } catch (ArrayIndexOutOfBoundsException e) {
+            throw JellyExceptions.rdfProtoDeserializationError(
+                "Encountered an invalid name table reference (out of bounds). " +
+                "Name ID: " +
+                nameId +
+                ", Prefix ID: " +
+                iri.prefixId()
+            );
+        }
         int prefixId = iri.prefixId();
         // Branchless way to update the prefix ID
         // Equivalent to:
@@ -117,7 +138,18 @@ public TIri decode(RdfIri iri) {
         lastPrefixIdReference = prefixId = (((prefixId - 1) >> 31) & lastPrefixIdReference) + prefixId;
         if (prefixId != 0) {
             // Name and prefix
-            PrefixLookupEntry prefixEntry = prefixLookup[prefixId];
+            PrefixLookupEntry prefixEntry;
+            try {
+                prefixEntry = prefixLookup[prefixId];
+            } catch (ArrayIndexOutOfBoundsException e) {
+                throw JellyExceptions.rdfProtoDeserializationError(
+                    "Encountered an invalid prefix table reference (out of bounds). " +
+                    "Prefix ID: " +
+                    prefixId +
+                    ", Name ID: " +
+                    nameId
+                );
+            }
             if (nameEntry.lastPrefixId != prefixId || nameEntry.lastPrefixSerial != prefixEntry.serial) {
                 // Update the last prefix
                 nameEntry.lastPrefixId = prefixId;

core/src/main/java/eu/ostrzyciel/jelly/core/internal/TranscoderLookup.java

@@ -1,5 +1,6 @@
 package eu.ostrzyciel.jelly.core.internal;
 
+import eu.ostrzyciel.jelly.core.JellyExceptions;
 import java.util.Arrays;
 
 /**
@@ -101,7 +102,9 @@ int remap(int id) {
      */
     void newInputStream(int size) {
         if (size > outputSize) {
-            throw new IllegalArgumentException("Input lookup size cannot be greater than the output lookup size");
+            throw JellyExceptions.rdfProtoTranscodingError(
+                "Input lookup size cannot be greater than the output lookup size"
+            );
         }
         if (table != null) {
             // Only set this for streams 2 and above (counting from 1)

core/src/main/scala/eu/ostrzyciel/jelly/core/JellyExceptions.scala

@@ -30,4 +30,12 @@ private object JellyExceptions extends JellyExceptions:
   private[core] def rdfProtoSerializationError(msg: String): RdfProtoSerializationError =
     new RdfProtoSerializationError(msg)
 
+  /**
+   * Helper method to allow Java code to throw a [[RdfProtoTranscodingError]].
+   * @param msg error message
+   * @return an instance of [[RdfProtoTranscodingError]]
+   */
+  private[core] def rdfProtoTranscodingError(msg: String): RdfProtoTranscodingError =
+    new RdfProtoTranscodingError(msg)
+
 export JellyExceptions.*

core/src/test/scala/eu/ostrzyciel/jelly/core/ProtoDecoderSpec.scala

@@ -272,22 +272,26 @@ class ProtoDecoderSpec extends AnyWordSpec, Matchers:
 
     // The tests for this logic are in internal.NameDecoderSpec
     // Here we are just testing if the exceptions are rethrown correctly.
-    "throw exception on out-of-bounds references to lookups" in {
+    "throw exception on an invalid IRI term" in {
       val decoder = MockConverterFactory.triplesDecoder(None)
       val data = wrapEncodedFull(Seq(
         JellyOptions.smallGeneralized.withPhysicalType(PhysicalStreamType.TRIPLES),
+        RdfPrefixEntry(0, null),
+        RdfNameEntry(0, null),
         RdfTriple(
           RdfTerm.Bnode("1"),
           RdfTerm.Bnode("2"),
-          RdfIri(10000, 0),
+          RdfIri(1, 1),
         ),
       ))
       decoder.ingestRow(data.head)
+      decoder.ingestRow(data(1))
+      decoder.ingestRow(data(2))
       val error = intercept[RdfProtoDeserializationError] {
-        decoder.ingestRow(data(1))
+        decoder.ingestRow(data(3))
       }
       error.getMessage should include ("Error while decoding term")
-      error.getCause shouldBe a [ArrayIndexOutOfBoundsException]
+      error.getCause shouldBe a [NullPointerException]
     }
   }
 
@@ -461,20 +465,28 @@ class ProtoDecoderSpec extends AnyWordSpec, Matchers:
 
     // The tests for this logic are in internal.NameDecoderSpec
     // Here we are just testing if the exceptions are rethrown correctly.
-    "throw exception on out-of-bounds references to lookups (graph term)" in {
-      val decoder = MockConverterFactory.graphsAsQuadsDecoder(None)
+    "throw exception on an invalid IRI term" in {
+      val decoder = MockConverterFactory.graphsAsQuadsDecoder()
       val data = wrapEncodedFull(Seq(
         JellyOptions.smallGeneralized.withPhysicalType(PhysicalStreamType.GRAPHS),
-        RdfGraphStart(
-          RdfIri(10000, 0),
+        RdfPrefixEntry(0, null),
+        RdfNameEntry(0, null),
+        RdfGraphStart(RdfDefaultGraph.defaultInstance),
+        RdfTriple(
+          RdfTerm.Bnode("1"),
+          RdfTerm.Bnode("2"),
+          RdfIri(1, 1),
         ),
       ))
       decoder.ingestRow(data.head)
+      decoder.ingestRow(data(1))
+      decoder.ingestRow(data(2))
+      decoder.ingestRow(data(3))
       val error = intercept[RdfProtoDeserializationError] {
-        decoder.ingestRow(data(1))
+        decoder.ingestRow(data(4))
       }
-      error.getMessage should include ("Error while decoding graph term")
-      error.getCause shouldBe a [ArrayIndexOutOfBoundsException]
+      error.getMessage should include("Error while decoding term")
+      error.getCause shouldBe a[NullPointerException]
     }
   }
 

core/src/test/scala/eu/ostrzyciel/jelly/core/internal/NameDecoderSpec.scala

@@ -103,66 +103,75 @@ class NameDecoderSpec extends AnyWordSpec, Matchers:
 
       "not accept a new prefix ID larger than table size" in {
         val dec = makeDecoder(smallOptions)
-        intercept[ArrayIndexOutOfBoundsException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.updatePrefixes(RdfPrefixEntry(9, "https://test.org/"))
         }
+        e.getMessage should include ("Prefix entry with ID 9")
       }
 
       "not accept a new prefix ID lower than 0 (-1)" in {
         val dec = makeDecoder(smallOptions)
-        intercept[NullPointerException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.updatePrefixes(RdfPrefixEntry(-1, "https://test.org/"))
         }
+        e.getMessage should include ("Prefix entry with ID -1")
       }
 
       "not accept a new prefix ID lower than 0 (-2)" in {
         val dec = makeDecoder(smallOptions)
-        intercept[ArrayIndexOutOfBoundsException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.updatePrefixes(RdfPrefixEntry(-2, "https://test.org/"))
         }
+        e.getMessage should include ("Prefix entry with ID -2")
       }
 
       "not retrieve a prefix ID larger than table size" in {
         val dec = makeDecoder(smallOptions)
-        intercept[ArrayIndexOutOfBoundsException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.decode(RdfIri(9, 0))
         }
+        e.getMessage should include ("Prefix ID: 9")
       }
 
       "not accept a new name ID larger than table size" in {
         val dec = makeDecoder(smallOptions)
-        intercept[ArrayIndexOutOfBoundsException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.updateNames(RdfNameEntry(17, "Cake"))
         }
+        e.getMessage should include ("Name entry with ID 17")
       }
 
       "not accept a default ID going beyond the table size" in {
         val dec = makeDecoder(smallOptions)
         dec.updateNames(RdfNameEntry(16, "Cake"))
-        intercept[ArrayIndexOutOfBoundsException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.updateNames(RdfNameEntry(0, "Cake 2"))
         }
+        e.getMessage should include ("Name entry with ID 0")
       }
 
       "not accept a new name ID lower than 0 (-1)" in {
         val dec = makeDecoder(smallOptions)
-        intercept[NullPointerException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.updateNames(RdfNameEntry(-1, "Cake"))
         }
+        e.getMessage should include ("Name entry with ID -1")
       }
 
       "not accept a new name ID lower than 0 (-2)" in {
         val dec = makeDecoder(smallOptions)
-        intercept[ArrayIndexOutOfBoundsException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.updateNames(RdfNameEntry(-2, "Cake"))
         }
+        e.getMessage should include ("Name entry with ID -2")
       }
 
       "not retrieve a name ID larger than table size" in {
         val dec = makeDecoder(smallOptions)
-        intercept[ArrayIndexOutOfBoundsException] {
+        val e = intercept[RdfProtoDeserializationError] {
           dec.decode(RdfIri(0, 17))
         }
+        e.getMessage should include ("Name ID: 17")
       }
     }
   }

core/src/test/scala/eu/ostrzyciel/jelly/core/internal/TranscoderLookupSpec.scala

@@ -1,5 +1,6 @@
 package eu.ostrzyciel.jelly.core.internal
 
+import eu.ostrzyciel.jelly.core.JellyExceptions.RdfProtoTranscodingError
 import org.scalatest.matchers.should.Matchers
 import org.scalatest.wordspec.AnyWordSpec
 
@@ -11,7 +12,7 @@ class TranscoderLookupSpec extends AnyWordSpec, Matchers:
   "TranscoderLookup" should {
     "throw an exception when trying to set input lookup size greater than the output" in {
       val tl = TranscoderLookup(false, 100)
-      val ex = intercept[IllegalArgumentException] {
+      val ex = intercept[RdfProtoTranscodingError] {
         tl.newInputStream(120)
       }
       ex.getMessage should include ("Input lookup size cannot be greater than the output lookup size")