Fix BASE resolution for formats that support it (#222)

Author: Ostrzyciel

src/main/scala/eu/neverblink/jelly/cli/command/rdf/RdfToJelly.scala

@@ -14,7 +14,7 @@ import eu.neverblink.jelly.core.proto.google.v1 as google
 import eu.neverblink.jelly.core.proto.v1.*
 import eu.neverblink.jelly.core.utils.IoUtils
 import org.apache.jena.riot.system.StreamRDFWriter
-import org.apache.jena.riot.{Lang, RIOT}
+import org.apache.jena.riot.RIOT
 
 import java.io.{BufferedReader, FileInputStream, InputStream, InputStreamReader, OutputStream}
 import scala.util.Using
@@ -77,7 +77,7 @@ object RdfToJelly extends RdfSerDesCommand[RdfToJellyOptions, RdfFormat.Readable
   lazy val printUtil: RdfCommandPrintUtil[RdfFormat.Readable] = RdfToJellyPrint
 
   val defaultAction: (InputStream, OutputStream) => Unit =
-    langToJelly(RdfFormat.NQuads.jenaLang, _, _)
+    langToJelly(RdfFormat.NQuads, _, _)
 
   private def loadOptionsFromFile(filename: String): RdfStreamOptions =
     val inputStream = new FileInputStream(filename)
@@ -114,20 +114,20 @@ object RdfToJelly extends RdfSerDesCommand[RdfToJellyOptions, RdfFormat.Readable
   override def matchFormatToAction(
       format: RdfFormat.Readable,
   ): Option[(InputStream, OutputStream) => Unit] = format match {
-    case f: RdfFormat.Jena.Readable => Some(langToJelly(f.jenaLang, _, _))
+    case f: RdfFormat.Jena.Readable => Some(langToJelly(f, _, _))
     case f: RdfFormat.JellyText.type => Some(jellyTextToJelly)
   }
 
   /** This method reads the file, rewrites it to Jelly and writes it to some output stream
-    * @param jenaLang
+    * @param format
     *   Language that should be converted to Jelly
     * @param inputStream
     *   InputStream
     * @param outputStream
     *   OutputStream
     */
   private def langToJelly(
-      jenaLang: Lang,
+      format: RdfFormat.Jena,
       inputStream: InputStream,
       outputStream: OutputStream,
   ): Unit =
@@ -189,8 +189,8 @@ object RdfToJelly extends RdfSerDesCommand[RdfToJellyOptions, RdfFormat.Readable
           JellyStreamWriter(JenaConverterFactory.getInstance(), variant, outputStream)
 
     RiotParserUtil.parse(
-      getOptions.rdfPerformanceOptions.validateTerms.getOrElse(false),
-      jenaLang,
+      getOptions.rdfPerformanceOptions.resolveIris,
+      format,
       inputStream,
       jellyWriter,
     )

src/main/scala/eu/neverblink/jelly/cli/command/rdf/RdfValidate.scala

@@ -250,8 +250,8 @@ object RdfValidate extends JellyCommand[RdfValidateOptions]:
     val output = StreamRdfCollector()
     Using.resource(IoUtil.inputStream(fileName)) { is =>
       RiotParserUtil.parse(
-        getOptions.rdfPerformanceOptions.validateTerms.getOrElse(true),
-        format.jenaLang,
+        getOptions.rdfPerformanceOptions.resolveIris,
+        format,
         is,
         output,
       )

src/main/scala/eu/neverblink/jelly/cli/command/rdf/util/RdfFormat.scala

@@ -6,6 +6,7 @@ import org.apache.jena.riot.{Lang, RDFLanguages}
 sealed trait RdfFormat:
   val fullName: String
   val cliOptions: List[String]
+  val supportsBaseIri: Boolean
 
 object RdfFormat:
 
@@ -29,6 +30,7 @@ object RdfFormat:
     override val fullName: String = "N-Quads"
     override val cliOptions: List[String] = List("nq", "nquads")
     override val jenaLang: Lang = RDFLanguages.NQUADS
+    override val supportsBaseIri: Boolean = false
 
   case object NTriples
       extends RdfFormat.Jena.StreamWriteable,
@@ -37,16 +39,19 @@ object RdfFormat:
     override val fullName: String = "N-Triples"
     override val cliOptions: List[String] = List("nt", "ntriples")
     override val jenaLang: Lang = RDFLanguages.NTRIPLES
+    override val supportsBaseIri: Boolean = false
 
   case object Turtle extends RdfFormat.Jena.StreamWriteable, RdfFormat.Jena.Readable:
     override val fullName: String = "Turtle"
     override val cliOptions: List[String] = List("ttl", "turtle")
     override val jenaLang: Lang = RDFLanguages.TURTLE
+    override val supportsBaseIri: Boolean = true
 
   case object TriG extends RdfFormat.Jena.StreamWriteable, RdfFormat.Jena.Readable:
     override val fullName: String = "TriG"
     override val cliOptions: List[String] = List("trig")
     override val jenaLang: Lang = RDFLanguages.TRIG
+    override val supportsBaseIri: Boolean = true
 
   case object RdfProto
       extends RdfFormat.Jena.StreamWriteable,
@@ -55,6 +60,7 @@ object RdfFormat:
     override val fullName: String = "RDF Protobuf"
     override val cliOptions: List[String] = List("jenaproto", "jena-proto")
     override val jenaLang: Lang = RDFLanguages.RDFPROTO
+    override val supportsBaseIri: Boolean = false
 
   case object Thrift
       extends RdfFormat.Jena.StreamWriteable,
@@ -63,23 +69,27 @@ object RdfFormat:
     override val fullName: String = "RDF Thrift"
     override val cliOptions: List[String] = List("jenathrift", "jena-thrift")
     override val jenaLang: Lang = RDFLanguages.RDFTHRIFT
+    override val supportsBaseIri: Boolean = false
 
   case object RdfXml extends RdfFormat.Jena.Readable, RdfFormat.Jena.BatchWriteable:
     override val fullName: String = "RDF/XML"
     override val cliOptions: List[String] = List("rdfxml", "rdf-xml")
     override val jenaLang: Lang = RDFLanguages.RDFXML
+    override val supportsBaseIri: Boolean = true
 
   case object JsonLd extends RdfFormat.Jena.Readable, RdfFormat.Jena.BatchWriteable:
     override val fullName: String = "JSON-LD"
     override val cliOptions: List[String] = List("jsonld", "json-ld")
     override val jenaLang: Lang = RDFLanguages.JSONLD
+    override val supportsBaseIri: Boolean = true
 
   // We do not ever want to write or read from Jelly to Jelly
   // So better not have it as Writeable or Readable, just mark that it's integrated into Jena
   case object JellyBinary extends RdfFormat.Jena, RdfFormat.SupportsGeneralizedRdf:
     override val fullName: String = "Jelly binary"
     override val cliOptions: List[String] = List("jelly")
     override val jenaLang: Lang = JellyLanguage.JELLY
+    override val supportsBaseIri: Boolean = false
 
   case object JellyText
       extends RdfFormat,
@@ -89,6 +99,7 @@ object RdfFormat:
     override val fullName: String = "Jelly text"
     override val cliOptions: List[String] = List("jelly-text")
     val extension = ".jelly.txt"
+    override val supportsBaseIri: Boolean = false
 
   private val rdfFormats: List[RdfFormat] =
     List(NQuads, NTriples, JellyBinary, JellyText, Turtle, TriG, RdfProto, Thrift, RdfXml, JsonLd)

src/main/scala/eu/neverblink/jelly/cli/command/rdf/util/RdfPerformanceOptions.scala

@@ -6,7 +6,14 @@ import caseapp.HelpMessage
   */
 case class RdfPerformanceOptions(
     @HelpMessage(
-      "Enable term validation and IRI resolution (slower). Default: false for all commands except 'rdf validate'.",
+      "Resolve IRIs with regard to the base specified in the input document. " +
+        "Disabling this will result in faster parsing of Turtle, JSON-LD and RDF/XML, but will " +
+        "also potentially result in relative IRIs in the output. " +
+        "Default: true (ignored for formats that don't support base IRIs).",
+    )
+    resolveIris: Boolean = true,
+    @HelpMessage(
+      "Enable term validation (slower). Default: false for all commands except 'rdf validate'.",
     )
     validateTerms: Option[Boolean] = None,
 )

src/main/scala/eu/neverblink/jelly/cli/util/jena/JenaSystemOptions.scala

@@ -1,12 +1,11 @@
 package eu.neverblink.jelly.cli.util.jena
 
 import org.apache.jena.graph.impl.LiteralLabel
-import org.apache.jena.irix.{IRIProviderAny, SystemIRIx}
 
 import scala.util.Try
 
 object JenaSystemOptions:
-  /** Enable faster parsing by disabling strict IRI and literal validation.
+  /** Enable faster parsing by disabling strict literal validation.
     * @return
     *   A Success if the operation was successful, or a Failure with the exception if not. The
     *   operation may fail in environments where reflection is not supported. The failure can be
@@ -21,13 +20,9 @@ object JenaSystemOptions:
     toggle(true)
 
   private def toggle(enable: Boolean): Try[Unit] =
-    val valueMode = if enable then
-      SystemIRIx.reset()
-      "EAGER"
-    else
-      // Set the IRI provider to one that does no validation or resolving whatsoever
-      SystemIRIx.setProvider(IRIProviderAny.stringProvider())
-      "LAZY"
+    val valueMode =
+      if enable then "EAGER"
+      else "LAZY"
 
     // Disable/enable eager computation of literal values, which does strict checking.
     // This requires reflection as the field is private static final.

src/main/scala/eu/neverblink/jelly/cli/util/jena/riot/RiotParserUtil.scala

@@ -1,6 +1,8 @@
 package eu.neverblink.jelly.cli.util.jena.riot
 
-import org.apache.jena.riot.{Lang, RDFParser, RDFParserRegistry, RIOT}
+import eu.neverblink.jelly.cli.command.rdf.util.RdfFormat
+import org.apache.jena.riot.lang.LabelToNode
+import org.apache.jena.riot.{RDFParser, RDFParserRegistry, RIOT}
 import org.apache.jena.riot.system.StreamRDF
 
 import java.io.InputStream
@@ -9,19 +11,24 @@ import java.io.InputStream
   */
 object RiotParserUtil:
   def parse(
-      enableTermValidation: Boolean,
-      lang: Lang,
+      resolveIris: Boolean,
+      format: RdfFormat.Jena,
       source: InputStream,
       output: StreamRDF,
-  ): Unit =
-    if enableTermValidation then
-      // Standard parser with validation enabled
+  ): Unit = {
+    // Only really enable IRI resolution if the format supports it
+    if resolveIris && format.supportsBaseIri then
+      // Parser with full IRI resolution
       RDFParser.source(source)
-        .lang(lang)
+        .lang(format.jenaLang)
+        .labelToNode(LabelToNode.createUseLabelAsGiven())
+        .checking(false)
+        .strict(false)
         .parse(output)
     else
       // Fast parser with validation disabled
       RDFParserRegistry
-        .getFactory(lang)
-        .create(lang, FastParserProfile())
-        .read(source, "", lang.getContentType, output, RIOT.getContext)
+        .getFactory(format.jenaLang)
+        .create(format.jenaLang, FastParserProfile())
+        .read(source, "", format.jenaLang.getContentType, output, RIOT.getContext)
+  }

src/test/scala/eu/neverblink/jelly/cli/command/rdf/RdfToJellySpec.scala

@@ -883,4 +883,56 @@ class RdfToJellySpec extends AnyWordSpec with TestFixtureHelper with Matchers:
           )
       }
     }
+
+    "handle IRI resolution" when {
+      "IRI resolution enabled (default), input TTL stream" in withEmptyJellyFile { j =>
+        val input =
+          """BASE <http://example.org/>
+            |<a> <http://example.org/p> <b> .
+            |""".stripMargin
+        RdfToJelly.setStdIn(ByteArrayInputStream(input.getBytes))
+        RdfToJelly.runTestCommand(
+          List("rdf", "to-jelly", "--in-format=ttl", "--to", j),
+        )
+        val content = translateJellyBack(new FileInputStream(j))
+        val stmts = content.listStatements().asScala.toSeq
+        stmts.size should be(1)
+        stmts.head.getSubject.getURI should be("http://example.org/a")
+        stmts.head.getPredicate.getURI should be("http://example.org/p")
+        stmts.head.getObject.asResource().getURI should be("http://example.org/b")
+      }
+
+      "IRI resolution disabled, input TTL stream" in withEmptyJellyFile { j =>
+        val input =
+          """BASE <http://example.org/>
+            |<a> <http://example.org/p> <b> .
+            |""".stripMargin
+        RdfToJelly.setStdIn(ByteArrayInputStream(input.getBytes))
+        RdfToJelly.runTestCommand(
+          List("rdf", "to-jelly", "--in-format=ttl", "--resolve-iris=false", "--to", j),
+        )
+        val content = translateJellyBack(new FileInputStream(j))
+        val stmts = content.listStatements().asScala.toSeq
+        stmts.size should be(1)
+        stmts.head.getSubject.getURI should be("a")
+        stmts.head.getPredicate.getURI should be("http://example.org/p")
+        stmts.head.getObject.asResource().getURI should be("b")
+      }
+
+      "IRI resolution enabled (but ignored), input NT stream" in withEmptyJellyFile { j =>
+        val input =
+          """<a> <http://example.org/p> <b> .
+            |""".stripMargin
+        RdfToJelly.setStdIn(ByteArrayInputStream(input.getBytes))
+        RdfToJelly.runTestCommand(
+          List("rdf", "to-jelly", "--to", j),
+        )
+        val content = translateJellyBack(new FileInputStream(j))
+        val stmts = content.listStatements().asScala.toSeq
+        stmts.size should be(1)
+        stmts.head.getSubject.getURI should be("a")
+        stmts.head.getPredicate.getURI should be("http://example.org/p")
+        stmts.head.getObject.asResource().getURI should be("b")
+      }
+    }
   }