[datasets] added THORAT for PYSRC and SEMGREP

.github/workflows/release.yaml

@@ -46,6 +46,7 @@ jobs:
           mv workspace/securibench-micro-JAVASRC.zip securibench-micro-JAVASRC.zip
           mv workspace/OWASP-BenchmarkJava-JAVA.zip OWASP-BenchmarkJava-JAVA.zip
           mv workspace/OWASP-BenchmarkJava-JAVASRC.zip OWASP-BenchmarkJava-JAVASRC.zip
+          mv workspace/THORAT.zip THORAT.zip
       - name: Set next release version
         id: taggerFinal
         uses: anothrNick/[email protected]
@@ -62,3 +63,4 @@ jobs:
             securibench-micro-JAVASRC.zip
             OWASP-BenchmarkJava-JAVA.zip
             OWASP-BenchmarkJava-JAVASRC.zip
+            THORAT.zip

src/main/scala/io/joern/benchmarks/datasets/BenchmarkDataset.scala

@@ -6,7 +6,8 @@ import io.joern.benchmarks.datasets.runner.{
   DatasetDownloader,
   IchnaeaDownloader,
   OWASPJavaDownloader,
-  SecuribenchMicroDownloader
+  SecuribenchMicroDownloader,
+  ThoratDownloader
 }
 import org.slf4j.LoggerFactory
 import upickle.default.*
@@ -50,7 +51,9 @@ object BenchmarkDataset {
       x => new SecuribenchMicroDownloader(x.datasetDir, JavaCpgTypes.SEMGREP)
     ),
     (AvailableBenchmarks.OWASP_SEMGREP, x => new OWASPJavaDownloader(x.datasetDir, JavaCpgTypes.SEMGREP)),
-    (AvailableBenchmarks.ICHNAEA_SEMGREP, x => new IchnaeaDownloader(x.datasetDir))
+    (AvailableBenchmarks.ICHNAEA_SEMGREP, x => new IchnaeaDownloader(x.datasetDir)),
+    (AvailableBenchmarks.THORAT_PYSRC, x => new ThoratDownloader(x.datasetDir)),
+    (AvailableBenchmarks.THORAT_SEMGREP, x => new ThoratDownloader(x.datasetDir))
   )
 
 }

src/main/scala/io/joern/benchmarks/datasets/runner/ThoratDownloader.scala

@@ -0,0 +1,40 @@
+package io.joern.benchmarks.datasets.runner
+
+import better.files.File
+import io.joern.benchmarks.*
+import org.slf4j.LoggerFactory
+import upickle.default.*
+
+import java.net.{URI, URL}
+import scala.util.{Failure, Success, Try}
+
+class ThoratDownloader(datasetDir: File) extends DatasetDownloader(datasetDir) with SingleFileDownloader {
+
+  private val logger = LoggerFactory.getLogger(getClass)
+
+  private val version        = "0.0.7"
+  override val benchmarkName = s"Thorat Python v$version"
+
+  override protected val benchmarkUrl: URL = URI(
+    s"https://github.com/DavidBakerEffendi/benchmark-for-taint-analysis-tools-for-python/archive/refs/tags/v$version.zip"
+  ).toURL
+  override protected val benchmarkFileName: String = s"benchmark-for-taint-analysis-tools-for-python-$version"
+  override protected val benchmarkBaseDir: File    = datasetDir / benchmarkFileName
+
+  override def initialize(): Try[File] = Try {
+    val outputFile = File(s"${datasetDir.pathAsString}/THORAT.zip")
+
+    if !outputFile.exists then
+      downloadBenchmarkAndUnarchive(CompressionTypes.ZIP)
+      compressBenchmark(benchmarkBaseDir, Option(File(s"${datasetDir.pathAsString}/THORAT.zip")))
+    else outputFile
+  }
+
+  override def run(): Unit = Try {
+    initialize() match {
+      case Failure(exception) =>
+        logger.error(s"Unable to initialize benchmark '$getClass'", exception)
+      case Success(benchmarkDir) =>
+    }
+  }
+}