generating (primitive) aggregated sarif

Author: nikololiahim

.gitignore

@@ -10,3 +10,4 @@ polystat_out/
 j2eo.jar
 sarif/**
 polystat-x86_64-pc-linux
+out/

.polystat.conf

@@ -1,11 +1,10 @@
 polystat {
-    lang = java
-    input = sandbox_java
+    lang = eo
+    input = sandbox
     tempDir = tmp
     outputFormats = [sarif]
     outputs = {
-        console = true,
         dirs = [out],
-        files = [./results/out.txt],
+        files = [all.json]
     }
 }

src/main/scala/org/polystat/EO.scala

@@ -8,6 +8,7 @@ import org.polystat.odin.analysis.EOOdinAnalyzer
 import org.polystat.odin.parser.EoParser.sourceCodeEoParser
 import cats.syntax.traverse.*
 import cats.syntax.foldable.*
+import io.circe.syntax.*
 
 object EO:
 
@@ -21,14 +22,25 @@ object EO:
 
   def analyze(cfg: ProcessedConfig): IO[Unit] =
     val inputFiles = readCodeFromInput(".eo", cfg.input)
-    inputFiles
+    val analyzed = inputFiles
+      .evalMap { case (codePath, code) =>
+        for analyzed <- runAnalyzers(cfg.filteredAnalyzers)(code)
+        yield (codePath, analyzed)
+      }
+      .compile
+      .toVector
+      .map(_.toMap)
+    val analyzeToDirs = inputFiles
       .evalMap { case (codePath, code) =>
         for
           _ <- IO.println(s"Analyzing $codePath...")
           analyzed <- runAnalyzers(cfg.filteredAnalyzers)(code)
           _ <- if cfg.output.console then IO.println(analyzed) else IO.unit
           _ <- cfg.fmts.traverse_ { case OutputFormat.Sarif =>
-            val sarifJson = SarifOutput(analyzed).json.toString
+            val sarifJson = SarifOutput(
+              codePath,
+              analyzed,
+            ).json.toString
             cfg.output.dirs.traverse_(out =>
               val outPath =
                 out / "sarif" / codePath.replaceExt(".sarif.json")
@@ -41,5 +53,21 @@ object EO:
       }
       .compile
       .drain
+    val analyzeAggregate = cfg.output.files.traverse_ { outputPath =>
+      cfg.fmts.traverse_ { case OutputFormat.Sarif =>
+        for
+          _ <- IO.println(s"Writing aggregated output to $outputPath...")
+          analyzed <- analyzed
+          sariOutput = AggregatedSarifOutput
+            .fromAnalyzed(analyzed)
+            .asJson
+            .deepDropNullValues
+            .toString
+          _ <- writeOutputTo(outputPath)(sariOutput)
+        yield ()
+      }
+    }
+
+    analyzeToDirs *> analyzeAggregate
   end analyze
 end EO

src/main/scala/org/polystat/Sarif.scala

@@ -5,20 +5,31 @@ import io.circe.Decoder
 import io.circe.Encoder
 import io.circe.Json
 import io.circe.syntax.*
+import fs2.io.file.Path
 import org.polystat.odin.analysis.EOOdinAnalyzer.OdinAnalysisResult
 import cats.syntax.foldable.*
+import cats.syntax.functorFilter.*
 
 import scala.CanEqual.derived
 
 import OdinAnalysisResult.*
 import Sarif.*
 
-final case class SarifOutput(errors: List[OdinAnalysisResult]):
+object AggregatedSarifOutput:
+  def fromAnalyzed(
+      analyzed: Map[Path, List[OdinAnalysisResult]]
+  ): Seq[SarifLog] =
+    analyzed.toSeq.map { case (path, results) =>
+      SarifOutput(path, results).sarif
+    }
+end AggregatedSarifOutput
+
+final case class SarifOutput(filePath: Path, errors: List[OdinAnalysisResult]):
   def json: Json = sarif.asJson.deepDropNullValues
 
   private val sarifRun: SarifRun = SarifRun(
     SarifTool(SarifDriver()),
-    results = errors.flatMap(sarifResult),
+    results = errors.mapFilter(sarifResult),
     invocations = errors.map(sarifInvocation),
   )
 
@@ -84,6 +95,13 @@ final case class SarifOutput(errors: List[OdinAnalysisResult]):
             level = SarifLevel.ERROR,
             kind = SarifKind.FAIL,
             message = SarifMessage(message.mkString_("\n")),
+            locations = Seq(
+              SarifLocation(physicalLocation =
+                SarifPhysicalLocation(artifactLocation =
+                  SarifArtifactLocation(uri = filePath.toNioPath.toUri.toString)
+                )
+              )
+            ),
           )
         )
       case Ok(ruleId) =>
@@ -93,11 +111,18 @@ final case class SarifOutput(errors: List[OdinAnalysisResult]):
             level = SarifLevel.NONE,
             kind = SarifKind.PASS,
             message = SarifMessage("No errors were found."),
+            locations = Seq(
+              SarifLocation(physicalLocation =
+                SarifPhysicalLocation(artifactLocation =
+                  SarifArtifactLocation(uri = filePath.toNioPath.toUri.toString)
+                )
+              )
+            ),
           )
         )
 end SarifOutput
 
-object Sarif extends App:
+object Sarif:
 
   final val SARIF_VERSION = "2.1.0"
   final val SARIF_SCHEMA =
@@ -128,6 +153,19 @@ object Sarif extends App:
       level: SarifLevel,
       kind: SarifKind,
       message: SarifMessage,
+      locations: Seq[SarifLocation],
+  ) derives Codec.AsObject
+
+  final case class SarifLocation(
+      physicalLocation: SarifPhysicalLocation
+  ) derives Codec.AsObject
+
+  final case class SarifPhysicalLocation(
+      artifactLocation: SarifArtifactLocation
+  ) derives Codec.AsObject
+
+  final case class SarifArtifactLocation(
+      uri: String
   ) derives Codec.AsObject
 
   enum SarifLevel: