Merge pull request #547 from eikek/joex-memory

Joex memory

build.sbt

@@ -446,7 +446,9 @@ val joex = project
     addCompilerPlugin(Dependencies.betterMonadicFor),
     buildInfoPackage := "docspell.joex",
     reStart / javaOptions ++= Seq(
-      s"-Dconfig.file=${(LocalRootProject / baseDirectory).value / "local" / "dev.conf"}"
+      s"-Dconfig.file=${(LocalRootProject / baseDirectory).value / "local" / "dev.conf"}",
+      "-Xmx1596M",
+      "-XX:+UseG1GC"
     ),
     Revolver.enableDebugging(port = 5051, suspend = false)
   )
@@ -494,7 +496,9 @@ val restserver = project
       (Compile / resourceDirectory).value.getParentFile / "templates"
     ),
     reStart / javaOptions ++= Seq(
-      s"-Dconfig.file=${(LocalRootProject / baseDirectory).value / "local" / "dev.conf"}"
+      s"-Dconfig.file=${(LocalRootProject / baseDirectory).value / "local" / "dev.conf"}",
+      "-Xmx150M",
+      "-XX:+UseG1GC"
     ),
     Revolver.enableDebugging(port = 5050, suspend = false)
   )

modules/analysis/src/main/scala/docspell/analysis/TextAnalyser.scala

@@ -31,9 +31,11 @@ object TextAnalyser {
       labels ++ dates.map(dl => dl.label.copy(label = dl.date.toString))
   }
 
-  def create[F[_]: Sync](cfg: TextAnalysisConfig): Resource[F, TextAnalyser[F]] =
+  def create[F[_]: Concurrent: Timer](
+      cfg: TextAnalysisConfig
+  ): Resource[F, TextAnalyser[F]] =
     Resource
-      .liftF(PipelineCache[F]())
+      .liftF(PipelineCache[F](cfg.clearStanfordPipelineInterval))
       .map(cache =>
         new TextAnalyser[F] {
           def annotate(

modules/analysis/src/main/scala/docspell/analysis/TextAnalysisConfig.scala

@@ -1,8 +1,10 @@
 package docspell.analysis
 
 import docspell.analysis.nlp.TextClassifierConfig
+import docspell.common._
 
 case class TextAnalysisConfig(
     maxLength: Int,
+    clearStanfordPipelineInterval: Duration,
     classifier: TextClassifierConfig
 )

modules/analysis/src/main/scala/docspell/analysis/nlp/PipelineCache.scala

@@ -1,5 +1,7 @@
 package docspell.analysis.nlp
 
+import scala.concurrent.duration.{Duration => _, _}
+
 import cats.Applicative
 import cats.effect._
 import cats.effect.concurrent.Ref
@@ -19,7 +21,7 @@ import org.log4s.getLogger
   */
 trait PipelineCache[F[_]] {
 
-  def obtain(key: String, settings: StanfordNerSettings): F[StanfordCoreNLP]
+  def obtain(key: String, settings: StanfordNerSettings): Resource[F, StanfordCoreNLP]
 
 }
 
@@ -28,20 +30,29 @@ object PipelineCache {
 
   def none[F[_]: Applicative]: PipelineCache[F] =
     new PipelineCache[F] {
-      def obtain(ignored: String, settings: StanfordNerSettings): F[StanfordCoreNLP] =
+      def obtain(
-        makeClassifier(settings).pure[F]
+          ignored: String,
+          settings: StanfordNerSettings
+      ): Resource[F, StanfordCoreNLP] =
+        Resource.liftF(makeClassifier(settings).pure[F])
     }
 
-  def apply[F[_]: Sync](): F[PipelineCache[F]] =
+  def apply[F[_]: Concurrent: Timer](clearInterval: Duration): F[PipelineCache[F]] =
-    Ref.of(Map.empty[String, Entry]).map(data => (new Impl[F](data): PipelineCache[F]))
-
-  final private class Impl[F[_]: Sync](data: Ref[F, Map[String, Entry]])
-      extends PipelineCache[F] {
-
-    def obtain(key: String, settings: StanfordNerSettings): F[StanfordCoreNLP] =
     for {
-        id  <- makeSettingsId(settings)
+      data       <- Ref.of(Map.empty[String, Entry])
-        nlp <- data.modify(cache => getOrCreate(key, id, cache, settings))
+      cacheClear <- CacheClearing.create(data, clearInterval)
+    } yield new Impl[F](data, cacheClear)
+
+  final private class Impl[F[_]: Sync](
+      data: Ref[F, Map[String, Entry]],
+      cacheClear: CacheClearing[F]
+  ) extends PipelineCache[F] {
+
+    def obtain(key: String, settings: StanfordNerSettings): Resource[F, StanfordCoreNLP] =
+      for {
+        _   <- cacheClear.withCache
+        id  <- Resource.liftF(makeSettingsId(settings))
+        nlp <- Resource.liftF(data.modify(cache => getOrCreate(key, id, cache, settings)))
       } yield nlp
 
     private def getOrCreate(
@@ -81,6 +92,79 @@ object PipelineCache {
     }
 
   }
+
+  trait CacheClearing[F[_]] {
+    def withCache: Resource[F, Unit]
+  }
+
+  object CacheClearing {
+    def none[F[_]: Applicative]: CacheClearing[F] =
+      new CacheClearing[F] {
+        def withCache: Resource[F, Unit] =
+          Resource.pure[F, Unit](())
+      }
+
+    def create[F[_]: Concurrent: Timer](
+        data: Ref[F, Map[String, Entry]],
+        interval: Duration
+    ): F[CacheClearing[F]] =
+      for {
+        counter  <- Ref.of(0L)
+        cleaning <- Ref.of(None: Option[Fiber[F, Unit]])
+        log = Logger.log4s(logger)
+        result <-
+          if (interval.millis <= 0)
+            log
+              .info("Disable clearing StanfordNLP cache, due to config setting")
+              .map(_ => none[F])
+          else
+            log
+              .info(s"Clearing StanfordNLP cache after $interval idle time")
+              .map(_ =>
+                new CacheClearingImpl[F](data, counter, cleaning, interval.toScala)
+              )
+      } yield result
+  }
+
+  final private class CacheClearingImpl[F[_]](
+      data: Ref[F, Map[String, Entry]],
+      counter: Ref[F, Long],
+      cleaningFiber: Ref[F, Option[Fiber[F, Unit]]],
+      clearInterval: FiniteDuration
+  )(implicit T: Timer[F], F: Concurrent[F])
+      extends CacheClearing[F] {
+    private[this] val log = Logger.log4s[F](logger)
+
+    def withCache: Resource[F, Unit] =
+      Resource.make(counter.update(_ + 1))(_ =>
+        counter.updateAndGet(_ - 1).flatMap(n => scheduleClearPipeline(n))
+      )
+
+    def scheduleClearPipeline(cnt: Long): F[Unit] =
+      if (cnt > 0) cancelClear
+      else cancelClear *> clearAllLater.flatMap(fiber => cleaningFiber.set(fiber.some))
+
+    private def cancelClear: F[Unit] =
+      cleaningFiber.getAndSet(None).flatMap {
+        case Some(fiber) => fiber.cancel *> logDontClear
+        case None        => ().pure[F]
+      }
+
+    private def clearAllLater: F[Fiber[F, Unit]] =
+      F.start(T.sleep(clearInterval) *> clearAll)
+
+    private def logDontClear: F[Unit] =
+      log.info("Cancel stanford cache clearing, as it has been used in between.")
+
+    def clearAll: F[Unit] =
+      log.info("Clearing stanford nlp cache now!") *>
+        data.set(Map.empty) *> Sync[F].delay {
+          // turns out that everything is cached in a static map
+          StanfordCoreNLP.clearAnnotatorPool()
+          System.gc();
+        }
+  }
+
   private def makeClassifier(settings: StanfordNerSettings): StanfordCoreNLP = {
     logger.info(s"Creating ${settings.lang.name} Stanford NLP NER classifier...")
     new StanfordCoreNLP(Properties.forSettings(settings))

modules/analysis/src/main/scala/docspell/analysis/nlp/StanfordNerClassifier.scala

@@ -1,12 +1,10 @@
 package docspell.analysis.nlp
 
-import scala.jdk.CollectionConverters._
-
 import cats.Applicative
-import cats.implicits._
 
+import scala.jdk.CollectionConverters._
+import cats.effect._
 import docspell.common._
-
 import edu.stanford.nlp.pipeline.{CoreDocument, StanfordCoreNLP}
 
 object StanfordNerClassifier {
@@ -22,13 +20,13 @@ object StanfordNerClassifier {
     * a new classifier must be created. It will then replace the
     * previous one.
     */
-  def nerAnnotate[F[_]: Applicative](
+  def nerAnnotate[F[_]: BracketThrow](
       cacheKey: String,
       cache: PipelineCache[F]
   )(settings: StanfordNerSettings, text: String): F[Vector[NerLabel]] =
     cache
       .obtain(cacheKey, settings)
-      .map(crf => runClassifier(crf, text))
+      .use(crf => Applicative[F].pure(runClassifier(crf, text)))
 
   def runClassifier(nerClassifier: StanfordCoreNLP, text: String): Vector[NerLabel] = {
     val doc = new CoreDocument(text)

modules/common/src/main/scala/docspell/common/ByteSize.scala

@@ -0,0 +1,69 @@
+package docspell.common
+
+import io.circe.Decoder
+import io.circe.Encoder
+
+final case class ByteSize(bytes: Long) {
+
+  def toHuman: String =
+    ByteSize.bytesToHuman(bytes)
+
+  def <=(other: ByteSize) =
+    bytes <= other.bytes
+
+  def >=(other: ByteSize) =
+    bytes >= other.bytes
+
+  def >(other: ByteSize) =
+    bytes > other.bytes
+
+  def -(other: ByteSize) =
+    ByteSize(bytes - other.bytes)
+
+  def +(other: ByteSize) =
+    ByteSize(bytes + other.bytes)
+}
+
+object ByteSize {
+
+  val zero = ByteSize(0L)
+
+  def bytesToHuman(bytes: Long): String =
+    if (math.abs(bytes) < 1024 && bytes != Long.MinValue) s"${bytes}B"
+    else {
+      val k = bytes / 1024.0
+      if (math.abs(k) < 1024) f"$k%.02fK"
+      else {
+        val m = k / 1024.0
+        if (math.abs(m) < 1024) f"$m%.02fM"
+        else f"${m / 1024.0}%.02fG"
+      }
+    }
+
+  def parse(str: String): Either[String, ByteSize] =
+    str.toLongOption
+      .map(ByteSize.apply)
+      .toRight(s"Not a valid size string: $str")
+      .orElse(span(str.toLowerCase) match {
+        case (num, "k") =>
+          Right(ByteSize(math.round(num.toDouble * 1024)))
+        case (num, "m") =>
+          Right(ByteSize(math.round(num.toDouble * 1024 * 1024)))
+        case (num, "g") =>
+          Right(ByteSize(math.round(num.toDouble * 1024 * 1024 * 1024)))
+        case _ =>
+          Left(s"Invalid byte string: $str")
+      })
+
+  private def span(str: String): (String, String) =
+    if (str.isEmpty) ("", "")
+    else (str.init, str.last.toString)
+
+  def unsafe(str: String): ByteSize =
+    parse(str).fold(sys.error, identity)
+
+  implicit val jsonDecoder: Decoder[ByteSize] =
+    Decoder.decodeLong.map(ByteSize.apply)
+  implicit val jsonEncoder: Encoder[ByteSize] =
+    Encoder.encodeLong.contramap(_.bytes)
+}

modules/common/src/main/scala/docspell/common/JvmInfo.scala

@@ -0,0 +1,103 @@
+package docspell.common
+
+import java.time.Instant
+
+import scala.jdk.CollectionConverters._
+
+import cats.effect._
+import cats.implicits._
+
+import io.circe.generic.semiauto.{deriveDecoder, deriveEncoder}
+import io.circe.{Decoder, Encoder}
+
+case class JvmInfo(
+    id: Ident,
+    pidHost: String,
+    ncpu: Int,
+    inputArgs: List[String],
+    libraryPath: String,
+    specVendor: String,
+    specVersion: String,
+    startTime: Timestamp,
+    uptime: Duration,
+    vmName: String,
+    vmVendor: String,
+    vmVersion: String,
+    heapUsage: JvmInfo.MemoryUsage,
+    props: Map[String, String]
+)
+
+object JvmInfo {
+
+  def create[F[_]: Sync](id: Ident): F[JvmInfo] =
+    MemoryUsage.createHeap[F].flatMap { mu =>
+      Sync[F].delay {
+        val rmb = management.ManagementFactory.getRuntimeMXBean()
+        val rt  = Runtime.getRuntime()
+        JvmInfo(
+          id,
+          pidHost = rmb.getName(),
+          ncpu = rt.availableProcessors(),
+          inputArgs = rmb.getInputArguments().asScala.toList,
+          libraryPath = rmb.getLibraryPath(),
+          specVendor = rmb.getSpecVendor(),
+          specVersion = rmb.getSpecVersion(),
+          startTime = Timestamp(Instant.ofEpochMilli(rmb.getStartTime())),
+          uptime = Duration.millis(rmb.getUptime()),
+          vmName = rmb.getVmName(),
+          vmVendor = rmb.getVmVendor(),
+          vmVersion = rmb.getVmVersion(),
+          heapUsage = mu,
+          props = rmb.getSystemProperties().asScala.toMap
+        )
+      }
+    }
+
+  case class MemoryUsage(
+      init: Long,
+      used: Long,
+      comitted: Long,
+      max: Long,
+      free: Long,
+      description: String
+  )
+
+  object MemoryUsage {
+
+    def apply(init: Long, used: Long, comitted: Long, max: Long): MemoryUsage = {
+      def str(n: Long) = ByteSize(n).toHuman
+
+      val free = max - used
+
+      val descr =
+        s"init=${str(init)}, used=${str(used)}, comitted=${str(comitted)}, max=${str(max)}, free=${str(free)}"
+      MemoryUsage(init, used, comitted, max, free, descr)
+    }
+
+    val empty = MemoryUsage(0, 0, 0, 0)
+
+    def createHeap[F[_]: Sync]: F[MemoryUsage] =
+      Sync[F].delay {
+        val mxb  = management.ManagementFactory.getMemoryMXBean()
+        val heap = mxb.getHeapMemoryUsage()
+        MemoryUsage(
+          init = math.max(0, heap.getInit()),
+          used = math.max(0, heap.getUsed()),
+          comitted = math.max(0, heap.getCommitted()),
+          max = math.max(0, heap.getMax())
+        )
+      }
+
+    implicit val jsonEncoder: Encoder[MemoryUsage] =
+      deriveEncoder[MemoryUsage]
+
+    implicit val jsonDecoder: Decoder[MemoryUsage] =
+      deriveDecoder[MemoryUsage]
+  }
+
+  implicit val jsonEncoder: Encoder[JvmInfo] =
+    deriveEncoder[JvmInfo]
+
+  implicit val jsonDecoder: Decoder[JvmInfo] =
+    deriveDecoder[JvmInfo]
+}

modules/joex/src/main/resources/reference.conf

@@ -276,6 +276,13 @@ docspell.joex {
     # files.
     working-dir = ${java.io.tmpdir}"/docspell-analysis"
 
+    # The StanfordCoreNLP library caches language models which
+    # requires quite some amount of memory. Setting this interval to a
+    # positive duration, the cache is cleared after this amount of
+    # idle time. Set it to 0 to disable it if you have enough memory,
+    # processing will be faster.
+    clear-stanford-nlp-interval = "15 minutes"
+
     regex-ner {
       # Whether to enable custom NER annotation. This uses the address
       # book of a collective as input for NER tagging (to automatically

modules/joex/src/main/scala/docspell/joex/Config.scala

@@ -60,6 +60,7 @@ object Config {
   case class TextAnalysis(
       maxLength: Int,
       workingDir: Path,
+      clearStanfordNlpInterval: Duration,
       regexNer: RegexNer,
       classification: Classification
   ) {
@@ -67,6 +68,7 @@ object Config {
     def textAnalysisConfig: TextAnalysisConfig =
       TextAnalysisConfig(
         maxLength,
+        clearStanfordNlpInterval,
         TextClassifierConfig(
           workingDir,
           NonEmptyList

modules/joex/src/main/scala/docspell/joex/JoexServer.scala

@@ -35,7 +35,7 @@ object JoexServer {
           .create[F](cfg, signal, pools.connectEC, pools.httpClientEC, pools.blocker)
 
       httpApp = Router(
-        "/api/info" -> InfoRoutes(),
+        "/api/info" -> InfoRoutes(cfg),
         "/api/v1"   -> JoexRoutes(joexApp)
       ).orNotFound
 

modules/joex/src/main/scala/docspell/joex/routes/InfoRoutes.scala

@@ -1,8 +1,10 @@
 package docspell.joex.routes
 
 import cats.effect.Sync
+import cats.implicits._
 
-import docspell.joex.BuildInfo
+import docspell.common.JvmInfo
+import docspell.joex.{BuildInfo, Config}
 import docspell.joexapi.model.VersionInfo
 
 import org.http4s.HttpRoutes
@@ -11,10 +13,11 @@ import org.http4s.dsl.Http4sDsl
 
 object InfoRoutes {
 
-  def apply[F[_]: Sync](): HttpRoutes[F] = {
+  def apply[F[_]: Sync](cfg: Config): HttpRoutes[F] = {
     val dsl = new Http4sDsl[F] {}
     import dsl._
-    HttpRoutes.of[F] { case GET -> (Root / "version") =>
+    HttpRoutes.of[F] {
+      case GET -> Root / "version" =>
         Ok(
           VersionInfo(
             BuildInfo.version,
@@ -24,6 +27,9 @@ object InfoRoutes {
             BuildInfo.gitDescribedVersion.getOrElse("")
           )
         )
+
+      case GET -> Root / "system" =>
+        JvmInfo.create[F](cfg.appId).flatMap(Ok(_))
     }
   }
 }

modules/restserver/src/main/scala/docspell/restserver/RestServer.scala

@@ -104,7 +104,8 @@ object RestServer {
   def adminRoutes[F[_]: Effect](cfg: Config, restApp: RestApp[F]): HttpRoutes[F] =
     Router(
       "fts"  -> FullTextIndexRoutes.admin(cfg, restApp.backend),
-      "user" -> UserRoutes.admin(restApp.backend)
+      "user" -> UserRoutes.admin(restApp.backend),
+      "info" -> InfoRoutes.admin(cfg)
     )
 
   def redirectTo[F[_]: Effect](path: String): HttpRoutes[F] = {

modules/restserver/src/main/scala/docspell/restserver/routes/InfoRoutes.scala

@@ -1,9 +1,11 @@
 package docspell.restserver.routes
 
 import cats.effect.Sync
+import cats.implicits._
 
+import docspell.common._
 import docspell.restapi.model.VersionInfo
-import docspell.restserver.BuildInfo
+import docspell.restserver.{BuildInfo, Config}
 
 import org.http4s.HttpRoutes
 import org.http4s.circe.CirceEntityEncoder._
@@ -26,4 +28,13 @@ object InfoRoutes {
       )
     }
   }
+
+  def admin[F[_]: Sync](cfg: Config): HttpRoutes[F] = {
+    val dsl = new Http4sDsl[F] {}
+    import dsl._
+    HttpRoutes.of[F] { case GET -> Root / "system" =>
+      JvmInfo.create[F](cfg.appId).flatMap(Ok(_))
+    }
+
+  }
 }

nix/module-joex.nix

@@ -118,6 +118,7 @@ let
         ];
       };
       working-dir = "/tmp/docspell-analysis";
+      clear-stanford-nlp-interval = "15 minutes";
     };
     processing = {
       max-due-date-years = 10;
@@ -771,6 +772,14 @@ in {
                 files.
               '';
             };
+            clear-stanford-nlp-interval = mkOption {
+              type = types.str;
+              default = defaults.text-analysis.clear-stanford-nlp-interval;
+              description = ''
+                Idle time after which the NLP caches are cleared to free
+                memory. If <= 0 clearing the cache is disabled.
+              '';
+            };
 
             regex-ner = mkOption {
               type = types.submodule({