Convert sourcepath to relative. (#385)

This also introduces a new config setting, `sourceRoot` that will need
to be set by the plugins using this to ensure that no matter where this
is executed from that the relativization process will work as expected.

Author: ckipp01

Diff for: scalac-scoverage-plugin/src/main/scala/scoverage/Serializer.scala

@@ -11,51 +11,74 @@ import scala.io.Source
 
 object Serializer {
 
+  val coverageDataFormatVersion = "3.0"
+
   // Write out coverage data to the given data directory, using the default coverage filename
-  def serialize(coverage: Coverage, dataDir: String): Unit =
-    serialize(coverage, coverageFile(dataDir))
+  def serialize(coverage: Coverage, dataDir: String, sourceRoot: String): Unit =
+    serialize(coverage, coverageFile(dataDir), new File(sourceRoot))
 
   // Write out coverage data to given file.
-  def serialize(coverage: Coverage, file: File): Unit = {
+  def serialize(coverage: Coverage, file: File, sourceRoot: File): Unit = {
     val writer: Writer = new BufferedWriter(
       new OutputStreamWriter(new FileOutputStream(file), Codec.UTF8.name)
     )
     try {
-      serialize(coverage, writer)
+      serialize(coverage, writer, sourceRoot)
     } finally {
       writer.flush()
       writer.close()
     }
   }
 
-  def serialize(coverage: Coverage, writer: Writer): Unit = {
+  def serialize(
+      coverage: Coverage,
+      writer: Writer,
+      sourceRoot: File
+  ): Unit = {
+    def getRelativePath(filePath: String): String = {
+      val base = sourceRoot.getCanonicalFile().toPath()
+      // NOTE: In the real world I have no idea if it's likely that you'll end
+      // up with weird issues on windows where the roots don't match, something
+      // like your root being D:/ and your file being C:/. If so this blows up.
+      // This happened on windows CI for me, since I was using a temp dir, and
+      // then trying to reletavize it off the cwd, which were in different
+      // drives. For now, we'll let this as is, but if 'other' has different
+      // root ever shows its, we'll shut that down real quick here... just not
+      // sure what to do in that situation yet.
+      val relPath =
+        base.relativize(new File(filePath).getCanonicalFile().toPath())
+      relPath.toString
+    }
+
     def writeHeader(writer: Writer): Unit = {
-      writer.write(s"""# Coverage data, format version: 2.0
-                      |# Statement data:
-                      |# - id
-                      |# - source path
-                      |# - package name
-                      |# - class name
-                      |# - class type (Class, Object or Trait)
-                      |# - full class name
-                      |# - method name
-                      |# - start offset
-                      |# - end offset
-                      |# - line number
-                      |# - symbol name
-                      |# - tree name
-                      |# - is branch
-                      |# - invocations count
-                      |# - is ignored
-                      |# - description (can be multi-line)
-                      |# '\f' sign
-                      |# ------------------------------------------
-                      |""".stripMargin.replaceAll("(\r\n)|\n|\r", "\n"))
+      writer.write(
+        s"""# Coverage data, format version: $coverageDataFormatVersion
+           |# Statement data:
+           |# - id
+           |# - source path
+           |# - package name
+           |# - class name
+           |# - class type (Class, Object or Trait)
+           |# - full class name
+           |# - method name
+           |# - start offset
+           |# - end offset
+           |# - line number
+           |# - symbol name
+           |# - tree name
+           |# - is branch
+           |# - invocations count
+           |# - is ignored
+           |# - description (can be multi-line)
+           |# '\f' sign
+           |# ------------------------------------------
+           |""".stripMargin
+      )
     }
 
     def writeStatement(stmt: Statement, writer: Writer): Unit = {
       writer.write(s"""${stmt.id}
-                      |${stmt.location.sourcePath}
+                      |${getRelativePath(stmt.location.sourcePath)}
                       |${stmt.location.packageName}
                       |${stmt.location.className}
                       |${stmt.location.classType}
@@ -71,7 +94,7 @@ object Serializer {
                       |${stmt.ignored}
                       |${stmt.desc}
                       |\f
-                      |""".stripMargin.replaceAll("(\r\n)|\n|\r", "\n"))
+                      |""".stripMargin)
     }
 
     writeHeader(writer)
@@ -84,13 +107,20 @@ object Serializer {
   def coverageFile(dataDir: String): File =
     new File(dataDir, Constants.CoverageFileName)
 
-  def deserialize(file: File): Coverage = {
+  def deserialize(file: File, sourceRoot: File): Coverage = {
     val source = Source.fromFile(file)(Codec.UTF8)
-    try deserialize(source.getLines())
+    try deserialize(source.getLines(), sourceRoot)
     finally source.close()
   }
 
-  def deserialize(lines: Iterator[String]): Coverage = {
+  def deserialize(lines: Iterator[String], sourceRoot: File): Coverage = {
+    // To integrate it smoothly with rest of the report writers,
+    // it is necessary to again convert [sourcePath] into a
+    // canonical one.
+    def getAbsolutePath(filePath: String): String = {
+      new File(sourceRoot, filePath).getCanonicalPath()
+    }
+
     def toStatement(lines: Iterator[String]): Statement = {
       val id: Int = lines.next().toInt
       val sourcePath = lines.next()
@@ -105,7 +135,7 @@ object Serializer {
         fullClassName,
         ClassType.fromString(classType),
         method,
-        sourcePath
+        getAbsolutePath(sourcePath)
       )
       val start: Int = lines.next().toInt
       val end: Int = lines.next().toInt
@@ -133,7 +163,7 @@ object Serializer {
 
     val headerFirstLine = lines.next()
     require(
-      headerFirstLine == "# Coverage data, format version: 2.0",
+      headerFirstLine == s"# Coverage data, format version: $coverageDataFormatVersion",
       "Wrong file format"
     )
 

Diff for: scalac-scoverage-plugin/src/main/scala/scoverage/plugin.scala

@@ -51,6 +51,8 @@ class ScoveragePlugin(val global: Global) extends Plugin {
         options.excludedSymbols = parseExclusionEntry("excludedSymbols:", opt)
       } else if (opt.startsWith("dataDir:")) {
         options.dataDir = opt.substring("dataDir:".length)
+      } else if (opt.startsWith("sourceRoot:")) {
+        options.sourceRoot = opt.substring("sourceRoot:".length())
       } else if (
         opt
           .startsWith("extraAfterPhase:") || opt.startsWith("extraBeforePhase:")
@@ -66,13 +68,18 @@ class ScoveragePlugin(val global: Global) extends Plugin {
       throw new RuntimeException(
         "Cannot invoke plugin without specifying <dataDir>"
       )
+    if (!opts.exists(_.startsWith("sourceRoot:")))
+      throw new RuntimeException(
+        "Cannot invoke plugin without specifying <sourceRoot>"
+      )
     instrumentationComponent.setOptions(options)
     true
   }
 
   override val optionsHelp: Option[String] = Some(
     Seq(
       "-P:scoverage:dataDir:<pathtodatadir>                  where the coverage files should be written\n",
+      "-P:scoverage:sourceRoot:<pathtosourceRoot>            the root dir of your sources, used for path relativization\n",
       "-P:scoverage:excludedPackages:<regex>;<regex>         semicolon separated list of regexs for packages to exclude",
       "-P:scoverage:excludedFiles:<regex>;<regex>            semicolon separated list of regexs for paths to exclude",
       "-P:scoverage:excludedSymbols:<regex>;<regex>          semicolon separated list of regexs for symbols to exclude",
@@ -107,6 +114,8 @@ class ScoveragePlugin(val global: Global) extends Plugin {
   }
 }
 
+// TODO refactor this into a case class. We'll also refactor how we parse the
+// options to get rid of all these vars
 class ScoverageOptions {
   var excludedPackages: Seq[String] = Nil
   var excludedFiles: Seq[String] = Nil
@@ -117,6 +126,12 @@ class ScoverageOptions {
   )
   var dataDir: String = IOUtils.getTempPath
   var reportTestName: Boolean = false
+  // TODO again, we'll refactor this later so this won't have a default here.
+  // However for tests we'll have to create this. However, make sure you create
+  // either both in temp or neither in temp, since on windows your temp dir
+  // will be in another drive, so the relativize functinality won't work if
+  // correctly.
+  var sourceRoot: String = IOUtils.getTempPath
 }
 
 class ScoverageInstrumentationComponent(
@@ -179,7 +194,12 @@ class ScoverageInstrumentationComponent(
         s"Instrumentation completed [${coverage.statements.size} statements]"
       )
 
-      Serializer.serialize(coverage, Serializer.coverageFile(options.dataDir))
+      // TODO do we need to verify this sourceRoot exists? How does semanticdb do this?
+      Serializer.serialize(
+        coverage,
+        Serializer.coverageFile(options.dataDir),
+        new File(options.sourceRoot)
+      )
       reporter.echo(
         s"Wrote instrumentation file [${Serializer.coverageFile(options.dataDir)}]"
       )

Diff for: scalac-scoverage-plugin/src/main/scala/scoverage/report/CoverageAggregator.scala

@@ -8,34 +8,32 @@ import scoverage.Serializer
 
 object CoverageAggregator {
 
-  @deprecated("1.4.0", "Used only by gradle-scoverage plugin")
-  def aggregate(baseDir: File, clean: Boolean): Option[Coverage] = {
-    aggregate(IOUtils.scoverageDataDirsSearch(baseDir))
-  }
-
   // to be used by gradle-scoverage plugin
-  def aggregate(dataDirs: Array[File]): Option[Coverage] = aggregate(
-    dataDirs.toSeq
-  )
+  def aggregate(dataDirs: Array[File], sourceRoot: File): Option[Coverage] =
+    aggregate(
+      dataDirs.toSeq,
+      sourceRoot
+    )
 
-  def aggregate(dataDirs: Seq[File]): Option[Coverage] = {
+  def aggregate(dataDirs: Seq[File], sourceRoot: File): Option[Coverage] = {
     println(
       s"[info] Found ${dataDirs.size} subproject scoverage data directories [${dataDirs.mkString(",")}]"
     )
     if (dataDirs.size > 0) {
-      Some(aggregatedCoverage(dataDirs))
+      Some(aggregatedCoverage(dataDirs, sourceRoot))
     } else {
       None
     }
   }
 
-  def aggregatedCoverage(dataDirs: Seq[File]): Coverage = {
+  def aggregatedCoverage(dataDirs: Seq[File], sourceRoot: File): Coverage = {
     var id = 0
     val coverage = Coverage()
     dataDirs foreach { dataDir =>
       val coverageFile: File = Serializer.coverageFile(dataDir)
       if (coverageFile.exists) {
-        val subcoverage: Coverage = Serializer.deserialize(coverageFile)
+        val subcoverage: Coverage =
+          Serializer.deserialize(coverageFile, sourceRoot)
         val measurementFiles: Array[File] =
           IOUtils.findMeasurementFiles(dataDir)
         val measurements = IOUtils.invoked(measurementFiles.toIndexedSeq)

Diff for: scalac-scoverage-plugin/src/test/scala/scoverage/CoverageAggregatorTest.scala

@@ -11,9 +11,9 @@ import scoverage.report.CoverageAggregator
 class CoverageAggregatorTest extends AnyFreeSpec with Matchers {
 
   // Let current directory be our source root
-  private val sourceRoot = new File(".")
+  private val sourceRoot = new File(".").getCanonicalPath()
   private def canonicalPath(fileName: String) =
-    new File(sourceRoot, fileName).getCanonicalPath
+    new File(sourceRoot, fileName).getCanonicalPath()
 
   "coverage aggregator" - {
     "should merge coverage objects with same id" in {
@@ -35,7 +35,11 @@ class CoverageAggregatorTest extends AnyFreeSpec with Matchers {
       coverage1.add(cov1Stmt2.copy(count = 0))
       val dir1 = new File(IOUtils.getTempPath, UUID.randomUUID.toString)
       dir1.mkdir()
-      Serializer.serialize(coverage1, Serializer.coverageFile(dir1))
+      Serializer.serialize(
+        coverage1,
+        Serializer.coverageFile(dir1),
+        new File(sourceRoot)
+      )
       val measurementsFile1 =
         new File(dir1, s"${Constants.MeasurementsPrefix}1")
       val measurementsFile1Writer = new FileWriter(measurementsFile1)
@@ -47,23 +51,34 @@ class CoverageAggregatorTest extends AnyFreeSpec with Matchers {
       coverage2.add(cov2Stmt1)
       val dir2 = new File(IOUtils.getTempPath, UUID.randomUUID.toString)
       dir2.mkdir()
-      Serializer.serialize(coverage2, Serializer.coverageFile(dir2))
+      Serializer.serialize(
+        coverage2,
+        Serializer.coverageFile(dir2),
+        new File(sourceRoot)
+      )
 
       val cov3Stmt1 =
         Statement(location, 2, 14, 1515, 544, "", "", "", false, 1)
       val coverage3 = Coverage()
       coverage3.add(cov3Stmt1.copy(count = 0))
       val dir3 = new File(IOUtils.getTempPath, UUID.randomUUID.toString)
       dir3.mkdir()
-      Serializer.serialize(coverage3, Serializer.coverageFile(dir3))
+      Serializer.serialize(
+        coverage3,
+        Serializer.coverageFile(dir3),
+        new File(sourceRoot)
+      )
       val measurementsFile3 =
         new File(dir3, s"${Constants.MeasurementsPrefix}1")
       val measurementsFile3Writer = new FileWriter(measurementsFile3)
       measurementsFile3Writer.write("2\n")
       measurementsFile3Writer.close()
 
       val aggregated =
-        CoverageAggregator.aggregatedCoverage(Seq(dir1, dir2, dir3))
+        CoverageAggregator.aggregatedCoverage(
+          Seq(dir1, dir2, dir3),
+          new File(sourceRoot)
+        )
       aggregated.statements.toSet.size shouldBe 4
       aggregated.statements.map(_.copy(id = 0)).toSet shouldBe
         Set(cov1Stmt1, cov1Stmt2, cov2Stmt1, cov3Stmt1).map(_.copy(id = 0))

Diff for: scalac-scoverage-plugin/src/test/scala/scoverage/ScoverageCompiler.scala

@@ -119,6 +119,7 @@ class ScoverageCompiler(
 
   val instrumentationComponent =
     new ScoverageInstrumentationComponent(this, None, None)
+
   instrumentationComponent.setOptions(new ScoverageOptions())
   val testStore = new ScoverageTestStoreComponent(this)
   val validator = new PositionValidator(this)