Allow configuring stanford-ner and cache based on collective

2025-06-21 18:08:25 +00:00 · 2020-08-24 00:56:25 +02:00
parent 4e7c00c345
commit 8628a0a8b3
11 changed files with 271 additions and 117 deletions
--- a/modules/joex/src/main/scala/docspell/joex/JoexAppImpl.scala
+++ b/modules/joex/src/main/scala/docspell/joex/JoexAppImpl.scala
@ -6,6 +6,7 @@ import cats.effect._
 import cats.implicits._
 import fs2.concurrent.SignallingRef

+import docspell.analysis.TextAnalyser
 import docspell.backend.ops._
 import docspell.common._
 import docspell.ftsclient.FtsClient
@ -80,14 +81,15 @@ object JoexAppImpl {
    for {
      httpClient <- BlazeClientBuilder[F](clientEC).resource
      client = JoexClient(httpClient)
-      store   <- Store.create(cfg.jdbc, connectEC, blocker)
-      queue   <- JobQueue(store)
-      pstore  <- PeriodicTaskStore.create(store)
-      nodeOps <- ONode(store)
-      joex    <- OJoex(client, store)
-      upload  <- OUpload(store, queue, cfg.files, joex)
-      fts     <- createFtsClient(cfg)(httpClient)
-      itemOps <- OItem(store, fts, queue, joex)
+      store    <- Store.create(cfg.jdbc, connectEC, blocker)
+      queue    <- JobQueue(store)
+      pstore   <- PeriodicTaskStore.create(store)
+      nodeOps  <- ONode(store)
+      joex     <- OJoex(client, store)
+      upload   <- OUpload(store, queue, cfg.files, joex)
+      fts      <- createFtsClient(cfg)(httpClient)
+      itemOps  <- OItem(store, fts, queue, joex)
+      analyser <- TextAnalyser.create[F](cfg.textAnalysis)
      javaEmil =
        JavaMailEmil(blocker, Settings.defaultSettings.copy(debug = cfg.mailDebug))
      sch <- SchedulerBuilder(cfg.scheduler, blocker, store)
@ -95,14 +97,14 @@ object JoexAppImpl {
        .withTask(
          JobTask.json(
            ProcessItemArgs.taskName,
-            ItemHandler.newItem[F](cfg, itemOps, fts),
+            ItemHandler.newItem[F](cfg, itemOps, fts, analyser),
            ItemHandler.onCancel[F]
          )
        )
        .withTask(
          JobTask.json(
            ReProcessItemArgs.taskName,
-            ReProcessItem[F](cfg, fts),
+            ReProcessItem[F](cfg, fts, analyser),
            ReProcessItem.onCancel[F]
          )
        )
--- a/modules/joex/src/main/scala/docspell/joex/process/ItemHandler.scala
+++ b/modules/joex/src/main/scala/docspell/joex/process/ItemHandler.scala
@ -5,6 +5,7 @@ import cats.effect._
 import cats.implicits._
 import fs2.Stream

+import docspell.analysis.TextAnalyser
 import docspell.backend.ops.OItem
 import docspell.common.{ItemState, ProcessItemArgs}
 import docspell.ftsclient.FtsClient
@ -29,11 +30,12 @@ object ItemHandler {
  def newItem[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
      itemOps: OItem[F],
-      fts: FtsClient[F]
+      fts: FtsClient[F],
+      analyser: TextAnalyser[F]
  ): Task[F, Args, Unit] =
    CreateItem[F]
      .flatMap(itemStateTask(ItemState.Processing))
-      .flatMap(safeProcess[F](cfg, itemOps, fts))
+      .flatMap(safeProcess[F](cfg, itemOps, fts, analyser))
      .map(_ => ())

  def itemStateTask[F[_]: Sync, A](
@ -51,11 +53,12 @@ object ItemHandler {
  def safeProcess[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
      itemOps: OItem[F],
-      fts: FtsClient[F]
+      fts: FtsClient[F],
+      analyser: TextAnalyser[F]
  )(data: ItemData): Task[F, Args, ItemData] =
    isLastRetry[F].flatMap {
      case true =>
-        ProcessItem[F](cfg, itemOps, fts)(data).attempt.flatMap({
+        ProcessItem[F](cfg, itemOps, fts, analyser)(data).attempt.flatMap({
          case Right(d) =>
            Task.pure(d)
          case Left(ex) =>
@ -65,7 +68,8 @@ object ItemHandler {
              .andThen(_ => Sync[F].raiseError(ex))
        })
      case false =>
-        ProcessItem[F](cfg, itemOps, fts)(data).flatMap(itemStateTask(ItemState.Created))
+        ProcessItem[F](cfg, itemOps, fts, analyser)(data)
+          .flatMap(itemStateTask(ItemState.Created))
    }

  private def markItemCreated[F[_]: Sync]: Task[F, Args, Boolean] =
--- a/modules/joex/src/main/scala/docspell/joex/process/ProcessItem.scala
+++ b/modules/joex/src/main/scala/docspell/joex/process/ProcessItem.scala
@ -2,6 +2,7 @@ package docspell.joex.process

 import cats.effect._

+import docspell.analysis.TextAnalyser
 import docspell.backend.ops.OItem
 import docspell.common.ProcessItemArgs
 import docspell.ftsclient.FtsClient
@ -13,25 +14,28 @@ object ProcessItem {
  def apply[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
      itemOps: OItem[F],
-      fts: FtsClient[F]
+      fts: FtsClient[F],
+      analyser: TextAnalyser[F]
  )(item: ItemData): Task[F, ProcessItemArgs, ItemData] =
    ExtractArchive(item)
      .flatMap(Task.setProgress(20))
-      .flatMap(processAttachments0(cfg, fts, (40, 60, 80)))
+      .flatMap(processAttachments0(cfg, fts, analyser, (40, 60, 80)))
      .flatMap(LinkProposal[F])
      .flatMap(SetGivenData[F](itemOps))
      .flatMap(Task.setProgress(99))

  def processAttachments[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
-      fts: FtsClient[F]
+      fts: FtsClient[F],
+      analyser: TextAnalyser[F]
  )(item: ItemData): Task[F, ProcessItemArgs, ItemData] =
-    processAttachments0[F](cfg, fts, (30, 60, 90))(item)
+    processAttachments0[F](cfg, fts, analyser, (30, 60, 90))(item)

  def analysisOnly[F[_]: Sync](
-      cfg: Config
+      cfg: Config,
+      analyser: TextAnalyser[F]
  )(item: ItemData): Task[F, ProcessItemArgs, ItemData] =
-    TextAnalysis[F](cfg.textAnalysis)(item)
+    TextAnalysis[F](analyser)(item)
      .flatMap(FindProposal[F](cfg.processing))
      .flatMap(EvalProposals[F])
      .flatMap(SaveProposals[F])
@ -39,12 +43,13 @@ object ProcessItem {
  private def processAttachments0[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
      fts: FtsClient[F],
+      analyser: TextAnalyser[F],
      progress: (Int, Int, Int)
  )(item: ItemData): Task[F, ProcessItemArgs, ItemData] =
    ConvertPdf(cfg.convert, item)
      .flatMap(Task.setProgress(progress._1))
      .flatMap(TextExtraction(cfg.extraction, fts))
      .flatMap(Task.setProgress(progress._2))
-      .flatMap(analysisOnly[F](cfg))
+      .flatMap(analysisOnly[F](cfg, analyser))
      .flatMap(Task.setProgress(progress._3))
 }
--- a/modules/joex/src/main/scala/docspell/joex/process/ReProcessItem.scala
+++ b/modules/joex/src/main/scala/docspell/joex/process/ReProcessItem.scala
@ -4,6 +4,7 @@ import cats.data.OptionT
 import cats.effect._
 import cats.implicits._

+import docspell.analysis.TextAnalyser
 import docspell.common._
 import docspell.ftsclient.FtsClient
 import docspell.joex.Config
@ -19,10 +20,11 @@ object ReProcessItem {

  def apply[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
-      fts: FtsClient[F]
+      fts: FtsClient[F],
+      analyser: TextAnalyser[F]
  ): Task[F, Args, Unit] =
    loadItem[F]
-      .flatMap(safeProcess[F](cfg, fts))
+      .flatMap(safeProcess[F](cfg, fts, analyser))
      .map(_ => ())

  def onCancel[F[_]: Sync: ContextShift]: Task[F, Args, Unit] =
@ -70,6 +72,7 @@ object ReProcessItem {
  def processFiles[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
      fts: FtsClient[F],
+      analyser: TextAnalyser[F],
      data: ItemData
  ): Task[F, Args, ItemData] = {

@ -91,7 +94,7 @@ object ReProcessItem {

    getLanguage[F].flatMap { lang =>
      ProcessItem
-        .processAttachments[F](cfg, fts)(data)
+        .processAttachments[F](cfg, fts, analyser)(data)
        .contramap[Args](convertArgs(lang))
    }
  }
@ -109,11 +112,12 @@ object ReProcessItem {

  def safeProcess[F[_]: ConcurrentEffect: ContextShift](
      cfg: Config,
-      fts: FtsClient[F]
+      fts: FtsClient[F],
+      analyser: TextAnalyser[F]
  )(data: ItemData): Task[F, Args, ItemData] =
    isLastRetry[F].flatMap {
      case true =>
-        processFiles[F](cfg, fts, data).attempt
+        processFiles[F](cfg, fts, analyser, data).attempt
          .flatMap({
            case Right(d) =>
              Task.pure(d)
@ -123,7 +127,7 @@ object ReProcessItem {
              ).andThen(_ => Sync[F].raiseError(ex))
          })
      case false =>
-        processFiles[F](cfg, fts, data)
+        processFiles[F](cfg, fts, analyser, data)
    }

  private def logWarn[F[_]](msg: => String): Task[F, Args, Unit] =
--- a/modules/joex/src/main/scala/docspell/joex/process/TextAnalysis.scala
+++ b/modules/joex/src/main/scala/docspell/joex/process/TextAnalysis.scala
@ -1,9 +1,10 @@
 package docspell.joex.process

-import cats.effect.Sync
+import cats.effect._
 import cats.implicits._

-import docspell.analysis.{TextAnalyser, TextAnalysisConfig}
+import docspell.analysis.TextAnalyser
+import docspell.analysis.nlp.StanfordSettings
 import docspell.common._
 import docspell.joex.process.ItemData.AttachmentDates
 import docspell.joex.scheduler.Task
@ -12,36 +13,40 @@ import docspell.store.records.RAttachmentMeta
 object TextAnalysis {

  def apply[F[_]: Sync](
-      cfg: TextAnalysisConfig
+      analyser: TextAnalyser[F]
  )(item: ItemData): Task[F, ProcessItemArgs, ItemData] =
    Task { ctx =>
-      TextAnalyser.create[F](cfg).use { analyser =>
-        for {
-          _ <- ctx.logger.info("Starting text analysis")
-          s <- Duration.stopTime[F]
-          t <-
-            item.metas.toList
-              .traverse(
-                annotateAttachment[F](ctx.args.meta.language, ctx.logger, analyser)
-              )
-          _ <- ctx.logger.debug(s"Storing tags: ${t.map(_._1.copy(content = None))}")
-          _ <- t.traverse(m =>
-            ctx.store.transact(RAttachmentMeta.updateLabels(m._1.id, m._1.nerlabels))
-          )
-          e <- s
-          _ <- ctx.logger.info(s"Text-Analysis finished in ${e.formatExact}")
-          v = t.toVector
-        } yield item.copy(metas = v.map(_._1), dateLabels = v.map(_._2))
-      }
+      for {
+        _ <- ctx.logger.info("Starting text analysis")
+        s <- Duration.stopTime[F]
+        t <-
+          item.metas.toList
+            .traverse(
+              annotateAttachment[F](ctx.args, ctx.logger, analyser)
+            )
+        _ <- ctx.logger.debug(s"Storing tags: ${t.map(_._1.copy(content = None))}")
+        _ <- t.traverse(m =>
+          ctx.store.transact(RAttachmentMeta.updateLabels(m._1.id, m._1.nerlabels))
+        )
+        e <- s
+        _ <- ctx.logger.info(s"Text-Analysis finished in ${e.formatExact}")
+        v = t.toVector
+      } yield item.copy(metas = v.map(_._1), dateLabels = v.map(_._2))
    }

  def annotateAttachment[F[_]: Sync](
-      lang: Language,
+      args: ProcessItemArgs,
      logger: Logger[F],
      analyser: TextAnalyser[F]
-  )(rm: RAttachmentMeta): F[(RAttachmentMeta, AttachmentDates)] =
+  )(rm: RAttachmentMeta): F[(RAttachmentMeta, AttachmentDates)] = {
+    val settings = StanfordSettings(args.meta.language, false, None)
    for {
-      labels <- analyser.annotate(logger, lang, rm.content.getOrElse(""))
+      labels <- analyser.annotate(
+        logger,
+        settings,
+        args.meta.collective,
+        rm.content.getOrElse("")
+      )
    } yield (rm.copy(nerlabels = labels.all.toList), AttachmentDates(rm, labels.dates))
-
+  }
 }