Use external binary to extract class name from stdin Java sources

Author: alexarchambault

build.sc

@@ -71,7 +71,6 @@ object `cli-options`  extends CliOptions
 object `build-macros` extends Cross[BuildMacros](Scala.mainVersions: _*)
 object options        extends Cross[Options](Scala.mainVersions: _*)
 object scalaparse     extends ScalaParse
-object javaparse      extends JavaParse
 object directives     extends Cross[Directives](Scala.mainVersions: _*)
 object core           extends Cross[Core](Scala.mainVersions: _*)
 object `build-module` extends Cross[Build](Scala.mainVersions: _*)
@@ -400,6 +399,7 @@ class Core(val crossScalaVersion: String) extends BuildLikeModule {
          |  def defaultGraalVMVersion = "${deps.graalVmVersion}"
          |
          |  def scalaCliSigningVersion = "${Deps.signingCli.dep.version}"
+         |  def javaClassNameVersion = "${Deps.javaClassName.dep.version}"
          |
          |  def libsodiumVersion = "${deps.libsodiumVersion}"
          |  def libsodiumjniVersion = "${Deps.libsodiumjni.dep.version}"
@@ -498,34 +498,6 @@ trait ScalaParse extends SbtModule with ScalaCliPublishModule with ScalaCliCompi
   def scalaVersion = Scala.scala213
 }
 
-trait JavaParse extends SbtModule with ScalaCliPublishModule with ScalaCliCompile {
-  def ivyDeps = super.ivyDeps() ++ Agg(Deps.scala3Compiler(scalaVersion()))
-
-  // pin scala3-library suffix, so that 2.13 modules can have us as moduleDep fine
-  def mandatoryIvyDeps = T {
-    super.mandatoryIvyDeps().map { dep =>
-      val isScala3Lib =
-        dep.dep.module.organization.value == "org.scala-lang" &&
-        dep.dep.module.name.value == "scala3-library" &&
-        (dep.cross match {
-          case _: CrossVersion.Binary => true
-          case _                      => false
-        })
-      if (isScala3Lib)
-        dep.copy(
-          dep = dep.dep.withModule(
-            dep.dep.module.withName(
-              coursier.ModuleName(dep.dep.module.name.value + "_3")
-            )
-          ),
-          cross = CrossVersion.empty(dep.cross.platformed)
-        )
-      else dep
-    }
-  }
-  def scalaVersion = Scala.scala3
-}
-
 trait Scala3Runtime extends SbtModule with ScalaCliPublishModule with ScalaCliCompile {
   def ivyDeps      = super.ivyDeps()
   def scalaVersion = Scala.scala3
@@ -561,7 +533,6 @@ class Build(val crossScalaVersion: String) extends BuildLikeModule {
   def moduleDeps = Seq(
     options(),
     scalaparse,
-    javaparse,
     directives(),
     `scala-cli-bsp`,
     `test-runner`(),
@@ -578,6 +549,7 @@ class Build(val crossScalaVersion: String) extends BuildLikeModule {
   def ivyDeps = super.ivyDeps() ++ Agg(
     Deps.asm,
     Deps.collectionCompat,
+    Deps.javaClassName,
     Deps.jsoniterCore,
     Deps.nativeTestRunner,
     Deps.osLib,

modules/build/src/main/java/scala/build/internal/JavaParserProxyMakerSubst.java

@@ -0,0 +1,22 @@
+package scala.build.internal;
+
+import com.oracle.svm.core.annotate.Substitute;
+import com.oracle.svm.core.annotate.TargetClass;
+
+/**
+ * This makes [[JavaParserProxyMaker.get]] provide a [[JavaParserProxyBinary]]
+ * rather than a [[JavaParserProxyJvm]], from native launchers.
+ *
+ * See [[JavaParserProxyMaker]] for more details.
+ */
+@TargetClass(className = "scala.build.internal.JavaParserProxyMaker")
+public final class JavaParserProxyMakerSubst {
+  @Substitute
+  public JavaParserProxy get(
+    Object archiveCache,
+    scala.Option<String> javaClassNameVersionOpt,
+    scala.build.Logger logger
+  ) {
+    return new JavaParserProxyBinary(archiveCache, logger, javaClassNameVersionOpt);
+  }
+}

modules/build/src/main/scala/scala/build/Build.scala

@@ -152,7 +152,9 @@ object Build {
       CrossSources.forInputs(
         inputs,
         Sources.defaultPreprocessors(
-          options.scriptOptions.codeWrapper.getOrElse(CustomCodeWrapper)
+          options.scriptOptions.codeWrapper.getOrElse(CustomCodeWrapper),
+          options.archiveCache,
+          options.internal.javaClassNameVersionOpt
         ),
         logger
       )

modules/build/src/main/scala/scala/build/Sources.scala

@@ -1,5 +1,8 @@
 package scala.build
 
+import coursier.cache.ArchiveCache
+import coursier.util.Task
+
 import scala.build.internal.CodeWrapper
 import scala.build.options.{BuildOptions, Scope}
 import scala.build.preprocessing.*
@@ -69,10 +72,26 @@ object Sources {
     topWrapperLen: Int
   )
 
-  def defaultPreprocessors(codeWrapper: CodeWrapper): Seq[Preprocessor] =
+  /** The default preprocessor list.
+    *
+    * @param codeWrapper
+    *   used by the Scala script preprocessor to "wrap" user code
+    * @param archiveCache
+    *   used from native launchers by the Java preprocessor, to download a java-class-name binary,
+    *   used to infer the class name of unnamed Java sources (like stdin)
+    * @param javaClassNameVersionOpt
+    *   if using a java-class-name binary, the version we should download. If empty, the default
+    *   version is downloaded.
+    * @return
+    */
+  def defaultPreprocessors(
+    codeWrapper: CodeWrapper,
+    archiveCache: ArchiveCache[Task],
+    javaClassNameVersionOpt: Option[String]
+  ): Seq[Preprocessor] =
     Seq(
       ScriptPreprocessor(codeWrapper),
-      JavaPreprocessor,
+      JavaPreprocessor(archiveCache, javaClassNameVersionOpt),
       ScalaPreprocessor,
       DataPreprocessor
     )

modules/build/src/main/scala/scala/build/bsp/BspImpl.scala

@@ -61,7 +61,9 @@ final class BspImpl(
       CrossSources.forInputs(
         inputs,
         Sources.defaultPreprocessors(
-          buildOptions.scriptOptions.codeWrapper.getOrElse(CustomCodeWrapper)
+          buildOptions.scriptOptions.codeWrapper.getOrElse(CustomCodeWrapper),
+          buildOptions.archiveCache,
+          buildOptions.internal.javaClassNameVersionOpt
         ),
         persistentLogger
       ).left.map((_, Scope.Main))

modules/build/src/main/scala/scala/build/internal/JavaParserProxy.scala

@@ -0,0 +1,22 @@
+package scala.build.internal
+
+import scala.build.errors.BuildException
+
+/** Helper to get class names from Java sources
+  *
+  * See [[JavaParserProxyJvm]] for the implementation that runs things in memory using
+  * java-class-name from the class path, and [[JavaParserProxyBinary]] for the implementation that
+  * downloads and runs a java-class-name binary.
+  */
+trait JavaParserProxy {
+
+  /** Extracts the class name of a Java source, using the dotty Java parser.
+    *
+    * @param content
+    *   the Java source to extract a class name from
+    * @return
+    *   either some class name (if one was found) or none (if none was found), or a
+    *   [[BuildException]]
+    */
+  def className(content: Array[Byte]): Either[BuildException, Option[String]]
+}

modules/build/src/main/scala/scala/build/internal/JavaParserProxyBinary.scala

@@ -0,0 +1,57 @@
+package scala.build.internal
+
+import coursier.cache.ArchiveCache
+import coursier.util.Task
+
+import scala.build.EitherCps.{either, value}
+import scala.build.Logger
+import scala.build.errors.BuildException
+import scala.util.Properties
+
+/** Downloads and runs java-class-name as an external binary. */
+class JavaParserProxyBinary(
+  archiveCache: ArchiveCache[Task],
+  javaClassNameVersionOpt: Option[String],
+  logger: Logger
+) extends JavaParserProxy {
+
+  /** For internal use only
+    *
+    * Passing archiveCache as an Object, to work around issues with higher-kind type params from
+    * Java code.
+    */
+  def this(
+    archiveCache: Object,
+    logger: Logger,
+    javaClassNameVersionOpt: Option[String]
+  ) =
+    this(archiveCache.asInstanceOf[ArchiveCache[Task]], javaClassNameVersionOpt, logger)
+
+  def className(content: Array[Byte]): Either[BuildException, Option[String]] = either {
+
+    val platformSuffix = FetchExternalBinary.platformSuffix()
+    val version        = javaClassNameVersionOpt.getOrElse(Constants.javaClassNameVersion)
+    val (tag, changing) =
+      if (version == "latest") ("nightly", true)
+      else ("v" + version, false)
+    val ext = if (Properties.isWin) ".zip" else ".gz"
+    val url =
+      s"https://github.com/scala-cli/java-class-name/releases/download/$tag/java-class-name-$platformSuffix$ext"
+
+    val binary =
+      value(FetchExternalBinary.fetch(url, changing, archiveCache, logger, "java-class-name"))
+
+    val source =
+      os.temp(content, suffix = ".java", perms = if (Properties.isWin) null else "rw-------")
+    val output =
+      try {
+        logger.debug(s"Running $binary $source")
+        val res = os.proc(binary, source).call()
+        res.out.text().trim
+      }
+      finally os.remove(source)
+
+    if (output.isEmpty) None
+    else Some(output)
+  }
+}

modules/build/src/main/scala/scala/build/internal/JavaParserProxyJvm.scala

@@ -0,0 +1,14 @@
+package scala.build.internal
+
+import scala.build.errors.BuildException
+import scala.cli.javaclassname.JavaParser
+
+/** A [[JavaParserProxy]] that relies on java-class-name in the class path, rather than downloading
+  * it and running it as an external binary.
+  *
+  * Should be used from Scala CLI when it's run on the JVM.
+  */
+class JavaParserProxyJvm extends JavaParserProxy {
+  override def className(content: Array[Byte]): Either[BuildException, Option[String]] =
+    Right(JavaParser.parseRootPublicClassName(content))
+}

modules/build/src/main/scala/scala/build/internal/JavaParserProxyMaker.scala

@@ -0,0 +1,24 @@
+package scala.build.internal
+
+import scala.build.Logger
+
+/** On the JVM, provides [[JavaParserProxyJvm]] as [[JavaParserProxy]] instance.
+  *
+  * From native launchers, [[JavaParserProxyMakerSubst]] takes over this, and gives
+  * [[JavaParserProxyBinary]] instead.
+  *
+  * That way, no reference to [[JavaParserProxyJvm]] remains in the native call graph, and that
+  * class and those it pulls (the java-class-name classes, which includes parts of the dotty parser)
+  * are not embedded the native launcher.
+  *
+  * Note that this is a class and not an object, to make it easier to write substitutions for that
+  * in Java.
+  */
+class JavaParserProxyMaker {
+  def get(
+    archiveCache: Object, // Actually a ArchiveCache[Task], but having issues with the higher-kind type param from Java…
+    javaClassNameVersionOpt: Option[String],
+    logger: Logger
+  ): JavaParserProxy =
+    new JavaParserProxyJvm
+}

modules/build/src/main/scala/scala/build/preprocessing/JavaPreprocessor.scala

@@ -1,18 +1,35 @@
 package scala.build.preprocessing
 
 import com.virtuslab.using_directives.custom.model.UsingDirectiveKind
+import coursier.cache.ArchiveCache
+import coursier.util.Task
 
 import java.nio.charset.StandardCharsets
 
 import scala.build.EitherCps.{either, value}
 import scala.build.errors.BuildException
-import scala.build.internal.JavaParser
+import scala.build.internal.JavaParserProxyMaker
 import scala.build.options.BuildRequirements
 import scala.build.preprocessing.ExtractedDirectives.from
 import scala.build.preprocessing.ScalaPreprocessor._
 import scala.build.{Inputs, Logger}
 
-case object JavaPreprocessor extends Preprocessor {
+/** Java source preprocessor.
+  *
+  * Doesn't modify Java sources. This only extracts using directives from them, and for unnamed
+  * sources (like stdin), tries to infer a class name from the sources themselves.
+  *
+  * @param archiveCache
+  *   when using a java-class-name external binary to infer a class name (see [[JavaParserProxy]]),
+  *   a cache to download that binary with
+  * @param javaClassNameVersionOpt
+  *   when using a java-class-name external binary to infer a class name (see [[JavaParserProxy]]),
+  *   this forces the java-class-name version to download
+  */
+final case class JavaPreprocessor(
+  archiveCache: ArchiveCache[Task],
+  javaClassNameVersionOpt: Option[String]
+) extends Preprocessor {
   def preprocess(
     input: Inputs.SingleElement,
     logger: Logger
@@ -45,27 +62,39 @@ case object JavaPreprocessor extends Preprocessor {
           ))
         })
       case v: Inputs.VirtualJavaFile =>
-        val relPath =
-          if (v.isStdin) {
-            val fileName = JavaParser.parseRootPublicClassName(v.content).map(
-              _ + ".java"
-            ).getOrElse("stdin.java")
-            os.sub / fileName
-          }
-          else v.subPath
-        val content = new String(v.content, StandardCharsets.UTF_8)
-        val s = PreprocessedSource.InMemory(
-          originalPath = Left(v.source),
-          relPath = relPath,
-          code = content,
-          ignoreLen = 0,
-          options = None,
-          requirements = None,
-          scopedRequirements = Nil,
-          mainClassOpt = None,
-          scopePath = v.scopePath
-        )
-        Some(Right(Seq(s)))
+        val res = either {
+          val relPath =
+            if (v.isStdin) {
+              val classNameOpt = value {
+                (new JavaParserProxyMaker)
+                  .get(
+                    archiveCache,
+                    javaClassNameVersionOpt,
+                    logger
+                  )
+                  .className(v.content)
+              }
+              val fileName = classNameOpt
+                .map(_ + ".java")
+                .getOrElse("stdin.java")
+              os.sub / fileName
+            }
+            else v.subPath
+          val content = new String(v.content, StandardCharsets.UTF_8)
+          val s = PreprocessedSource.InMemory(
+            originalPath = Left(v.source),
+            relPath = relPath,
+            code = content,
+            ignoreLen = 0,
+            options = None,
+            requirements = None,
+            scopedRequirements = Nil,
+            mainClassOpt = None,
+            scopePath = v.scopePath
+          )
+          Seq(s)
+        }
+        Some(res)
 
       case _ => None
     }