Merge pull request #1360 from adpi2/index-scala-2

Fix indexing of non-standard libs including scala-compiler, scala-library etc

modules/data/src/main/scala/scaladex/data/cleanup/NonStandardLib.scala

@@ -20,28 +20,30 @@ import scaladex.infra.DataPaths
 case class NonStandardLib(
     groupId: String,
     artifactId: String,
-    lookup: ScalaTargetLookup
+    lookup: BinaryVersionLookup
 )
 
-sealed trait ScalaTargetLookup
+sealed trait BinaryVersionLookup
 
-/**
- * The version is encoded in the pom file
- * dependency on org.scala-lang:scala-library
- * ex: io.gatling : gatling-compiler : 2.2.2
- */
-case object ScalaTargetFromPom extends ScalaTargetLookup
+object BinaryVersionLookup {
+  /**
+   * The version is encoded in the pom file
+   * dependency on org.scala-lang:scala-library
+   * ex: io.gatling : gatling-compiler : 2.2.2
+   */
+  case object FromDependency extends BinaryVersionLookup
 
-/**
- * The project is a plain-java project, thus no ScalaTarget.
- * ex: com.typesafe : config : 1.3.1
- */
-case object NoScalaTargetPureJavaDependency extends ScalaTargetLookup
+  /**
+   * The project is a plain-java project, thus no ScalaTarget.
+   * ex: com.typesafe : config : 1.3.1
+   */
+  case object Java extends BinaryVersionLookup
 
-/**
- * The version is encoded in the version (ex: scala-library itself)
- */
-case object ScalaTargetFromVersion extends ScalaTargetLookup
+  /**
+   * The version is encoded in the version (ex: scala-library itself)
+   */
+  case object FromArtifactVersion extends BinaryVersionLookup
+}
 
 object NonStandardLib {
 
@@ -58,9 +60,9 @@ object NonStandardLib {
         case (artifact, rawLookup) =>
           val lookup =
             rawLookup match {
-              case "pom"     => ScalaTargetFromPom
-              case "java"    => NoScalaTargetPureJavaDependency
-              case "version" => ScalaTargetFromVersion
+              case "pom"     => BinaryVersionLookup.FromDependency
+              case "java"    => BinaryVersionLookup.Java
+              case "version" => BinaryVersionLookup.FromArtifactVersion
               case _         => sys.error("unknown lookup: '" + rawLookup + "'")
             }
 

modules/server/src/main/scala/scaladex/server/Server.scala

@@ -69,7 +69,7 @@ object Server extends LazyLogging {
             val filesystem = FilesystemStorage(config.filesystem)
             val publishProcess = PublishProcess(paths, filesystem, webDatabase, config.env)(publishPool, system)
             val sonatypeClient = new SonatypeClientImpl()
-            val sonatypeSynchronizer = new SonatypeService(schedulerDatabase, sonatypeClient, publishProcess)
+            val sonatypeSynchronizer = new SonatypeService(paths, schedulerDatabase, sonatypeClient, publishProcess)
             val adminService =
               new AdminService(config.env, schedulerDatabase, searchEngine, githubClient, sonatypeSynchronizer)
 

modules/server/src/main/scala/scaladex/server/service/AdminService.scala

@@ -44,8 +44,14 @@ class AdminService(
       githubClientOpt.map { client =>
         val githubUpdater = new GithubUpdater(database, client)
         new JobScheduler(Job.githubInfo, githubUpdater.updateAll)
-      } ++
-      Option.when(!env.isLocal)(new JobScheduler(Job.missingMavenArtifacts, sonatypeSynchronizer.findMissing))
+      } ++ (
+        if (!env.isLocal) {
+          Seq(
+            new JobScheduler(Job.missingMavenArtifacts, sonatypeSynchronizer.findMissing),
+            new JobScheduler(Job.nonStandardArtifacts, sonatypeSynchronizer.findNonStandard)
+          )
+        } else Seq.empty
+      )
     seq.map(s => s.job.name -> s).toMap
   }
 

modules/server/src/main/scala/scaladex/server/service/ArtifactConverter.scala

@@ -68,14 +68,12 @@ class ArtifactConverter(paths: DataPaths) extends LazyLogging {
    * if the developer follow this convention we extract the relevant parts and we mark
    * the library as standard. Otherwise we either have a library like gatling or the scala library itself
    *
-   * @return The artifact name (without suffix), the Scala target, whether this project is a usual Scala library or not
+   * @return The artifact name (without suffix), the binary version, whether this project is a standard Scala library or not
    */
   private def extractMeta(pom: ArtifactModel): Option[ArtifactMeta] = {
     val nonStandardLookup =
-      nonStandardLibs
-        .find(lib =>
-          lib.groupId == pom.groupId &&
-            lib.artifactId == pom.artifactId
+      nonStandardLibs.find(lib =>
+          lib.groupId == pom.groupId && lib.artifactId == pom.artifactId
         )
         .map(_.lookup)
 
@@ -118,7 +116,7 @@ class ArtifactConverter(paths: DataPaths) extends LazyLogging {
         }
 
       // For example: io.gatling
-      case Some(ScalaTargetFromPom) =>
+      case Some(BinaryVersionLookup.FromDependency) =>
         for {
           dep <- pom.dependencies.find { dep =>
             dep.groupId == "org.scala-lang" &&
@@ -133,7 +131,7 @@ class ArtifactConverter(paths: DataPaths) extends LazyLogging {
           isNonStandard = true
         )
       // For example: typesafe config
-      case Some(NoScalaTargetPureJavaDependency) =>
+      case Some(BinaryVersionLookup.Java) =>
         Some(
           ArtifactMeta(
             artifactName = pom.artifactId,
@@ -143,7 +141,7 @@ class ArtifactConverter(paths: DataPaths) extends LazyLogging {
         )
 
       // For example: scala-compiler
-      case Some(ScalaTargetFromVersion) =>
+      case Some(BinaryVersionLookup.FromArtifactVersion) =>
         for (version <- SemanticVersion.parse(pom.version))
           yield ArtifactMeta(
             artifactName = pom.artifactId,

modules/server/src/main/scala/scaladex/server/service/SonatypeService.scala

@@ -9,49 +9,65 @@ import scaladex.core.model.Artifact._
 import scaladex.core.service.SchedulerDatabase
 import scaladex.core.service.SonatypeClient
 import scaladex.core.util.ScalaExtensions._
+import scaladex.data.cleanup.NonStandardLib
+import scaladex.infra.DataPaths
 
 class SonatypeService(
+    dataPaths: DataPaths,
     database: SchedulerDatabase,
     sonatypeService: SonatypeClient,
     publishProcess: PublishProcess
 )(implicit ec: ExecutionContext)
     extends LazyLogging {
-  import SonatypeService._
+
+  def findNonStandard(): Future[String] = {
+    val nonStandardLibs = NonStandardLib.load(dataPaths)
+    for {
+      mavenReferenceFromDatabase <- database.getAllMavenReferences()
+      result <- nonStandardLibs.mapSync { lib =>
+        val groupId = Artifact.GroupId(lib.groupId)
+        // get should not throw: it is a fixed set of artifactIds
+        val artifactId = Artifact.ArtifactId.parse(lib.artifactId).get 
+        findAndIndexMissingArtifacts(groupId, artifactId, mavenReferenceFromDatabase.toSet)
+      }
+    } yield s"Inserted ${result.sum} missing poms"
+  }
 
   def findMissing(): Future[String] =
     for {
-      groupIds <- database.getAllGroupIds()
+      mavenReferenceFromDatabase <- database.getAllMavenReferences().map(_.toSet)
+      groupIds = mavenReferenceFromDatabase.map(_.groupId).toSeq.sorted.map(Artifact.GroupId)
       // we sort just to estimate through the logs the percentage of progress
-      result <- groupIds.sortBy(_.value).mapSync(g => findAndIndexMissingArtifacts(g, None))
-    } yield s"Inserted ${result.size} missing poms"
+      result <- groupIds.mapSync(g => findAndIndexMissingArtifacts(g, None, mavenReferenceFromDatabase))
+    } yield s"Inserted ${result.sum} missing poms"
 
   def syncOne(groupId: GroupId, artifactNameOpt: Option[Artifact.Name]): Future[String] =
     for {
-      result <- findAndIndexMissingArtifacts(groupId, artifactNameOpt)
+      mavenReferenceFromDatabase <- database.getAllMavenReferences()
+      result <- findAndIndexMissingArtifacts(groupId, artifactNameOpt, mavenReferenceFromDatabase.toSet)
     } yield s"Inserted ${result} poms"
 
-  private def findAndIndexMissingArtifacts(groupId: GroupId, artifactNameOpt: Option[Artifact.Name]): Future[Int] =
+  private def findAndIndexMissingArtifacts(groupId: GroupId, artifactNameOpt: Option[Artifact.Name], knownRefs: Set[MavenReference]): Future[Int] =
     for {
-      mavenReferenceFromDatabase <- database.getAllMavenReferences()
       artifactIds <- sonatypeService.getAllArtifactIds(groupId)
       scalaArtifactIds = artifactIds.filter(artifact =>
         artifactNameOpt.forall(_ == artifact.name) && artifact.isScala && artifact.binaryVersion.isValid
       )
       result <- scalaArtifactIds
-        .mapSync(id => findAndIndexMissingArtifacts(groupId, id, mavenReferenceFromDatabase.toSet))
+        .mapSync(id => findAndIndexMissingArtifacts(groupId, id, knownRefs))
     } yield result.sum
 
   private def findAndIndexMissingArtifacts(
       groupId: GroupId,
       artifactId: ArtifactId,
-      mavenReferenceFromDatabase: Set[MavenReference]
+      knownRefs: Set[MavenReference]
   ): Future[Int] =
     for {
       versions <- sonatypeService.getAllVersions(groupId, artifactId)
       mavenReferences = versions.map(v =>
         MavenReference(groupId = groupId.value, artifactId = artifactId.value, version = v.toString)
       )
-      missingVersions = findMissingVersions(mavenReferenceFromDatabase, mavenReferences)
+      missingVersions = mavenReferences.filterNot(knownRefs)
       _ = if (missingVersions.nonEmpty)
         logger.warn(s"${missingVersions.size} artifacts are missing for ${groupId.value}:${artifactId.value}")
       missingPomFiles <- missingVersions.map(ref => sonatypeService.getPomFile(ref).map(_.map(ref -> _))).sequence
@@ -63,10 +79,4 @@ class SonatypeService(
       case PublishResult.Success => true
       case _                     => false
     }
-
-}
-
-object SonatypeService {
-  def findMissingVersions(fromDatabase: Set[MavenReference], fromSonatype: Seq[MavenReference]): Seq[MavenReference] =
-    fromSonatype.filterNot(fromDatabase)
 }

modules/template/src/main/scala/scaladex/view/Job.scala

@@ -44,6 +44,11 @@ object Job {
     "Find missing artifacts in Maven Central of the known group IDs.",
     24.hours
   )
+  val nonStandardArtifacts: Job = Job(
+    "non-standard-artifacts",
+    "Find missing non-standard artifacts from Maven Central",
+    2.hours
+  )
 
   case class Status(state: State, results: Seq[Result], progress: Option[Progress]) {
     def isStarted: Boolean = state.isInstanceOf[Started]