fathomnet/fathomnet-py#25 - first cut. Needs refinement

project/Dependencies.scala

@@ -9,18 +9,18 @@ object Dependencies {
 
   lazy val jansi    = "org.fusesource.jansi"         % "jansi"           % "2.4.1"
 
-  lazy val logback  = "ch.qos.logback"               % "logback-classic" % "1.4.14"
+  lazy val logback  = "ch.qos.logback"               % "logback-classic" % "1.5.1"
   lazy val methanol = "com.github.mizosoft.methanol" % "methanol"        % "1.7.0"
-  lazy val munit    = "org.scalameta"               %% "munit"           % "1.0.0-M10"
+  lazy val munit    = "org.scalameta"               %% "munit"           % "1.0.0-M11"
   lazy val picocli  = "info.picocli"                 % "picocli"         % "4.7.5"
 
-  lazy val slf4jJdk     = "org.slf4j"      % "slf4j-jdk-platform-logging"   % "2.0.10"
+  lazy val slf4jJdk     = "org.slf4j"      % "slf4j-jdk-platform-logging"   % "2.0.12"
 
-  private val tapirVersion  = "1.9.6"
+  private val tapirVersion  = "1.9.10"
   lazy val tapirStubServer  = "com.softwaremill.sttp.tapir" %% "tapir-sttp-stub-server" % tapirVersion
   lazy val tapirSwagger     = "com.softwaremill.sttp.tapir" %% "tapir-swagger-ui-bundle" % tapirVersion
   lazy val tapirCirce       = "com.softwaremill.sttp.tapir" %% "tapir-json-circe" % tapirVersion
-  lazy val tapirCirceClient = "com.softwaremill.sttp.client3" %% "circe" % "3.9.1"
+  lazy val tapirCirceClient = "com.softwaremill.sttp.client3" %% "circe" % "3.9.3"
   lazy val tapirNetty       = "com.softwaremill.sttp.tapir" %% "tapir-netty-server" % tapirVersion
   lazy val tapirVertx       = "com.softwaremill.sttp.tapir" %% "tapir-vertx-server" % tapirVersion
 

project/build.properties

@@ -1 +1 @@
-sbt.version=1.9.7
+sbt.version=1.9.9

src/main/scala/org/fathomnet/worms/Main.scala

@@ -127,7 +127,7 @@ object Main:
     WormsLoader.load(wormsDir).map { root =>
       if (treeFiles.nonEmpty)
         // Our new base. We use 0 as aphiaId so that the real aphiaIds are not incremented when the trees are combined
-        val newRoot      = WormsNode("object", "", 0L, Nil, Nil)
+        val newRoot      = WormsNode("object", "", 0L, 0L, Nil, Nil)
         val newBranches  = treeFiles.flatMap(ExtendedLoader.load(_)).toSeq
         val trees        = root +: newBranches
         val combinedRoot = CombineTrees.combine(newRoot, trees, root.maxAphiaId)

src/main/scala/org/fathomnet/worms/StateController.scala

@@ -9,6 +9,7 @@ package org.fathomnet.worms
 import scala.util.control.NonFatal
 import org.fathomnet.worms.etc.jdk.Logging.given
 import scala.collection.mutable.ListMap
+import cats.conversions.all
 
 object StateController:
 
@@ -62,6 +63,25 @@ object StateController:
       data.names.filter(_.toLowerCase.contains(glob.toLowerCase)).toList
     runSearch(search)
 
+  def findNamesByAphiaId(aphiaId: Long): Either[ErrorMsg, Names] =
+    def search(data: Data): Option[Names] =
+      val allNodes = data.namesMap.values
+      val existing = allNodes.find(n => n.aphiaId == aphiaId)
+      existing match
+        case None => None
+        case Some(node) =>
+          val accepted = if (node.aphiaId == node.acceptedAphiaId) 
+            Some(node)
+          else 
+            allNodes.find(n => n.aphiaId == node.acceptedAphiaId)
+          val names = accepted match
+            case None => Names(node.aphiaId, node.name, node.name, node.alternateNames)
+            case Some(value) => Names(node.aphiaId, node.name, value.name, node.alternateNames ++ value.alternateNames)
+
+          Option(names)
+
+    runNodeSearch(search, s"Unable to find a name with aphiaId: $aphiaId")
+
   def descendantNames(name: String): Either[ErrorMsg, List[String]] =
     def search(data: Data): List[String] =
       data.findNodeByName(name) match

src/main/scala/org/fathomnet/worms/WormsNode.scala

@@ -7,6 +7,7 @@
 package org.fathomnet.worms
 
 import org.fathomnet.worms.io.MutableWormsNode
+import org.fathomnet.worms.io.extended.CombineTrees.add
 
 /**
  * Immutable tree node
@@ -19,6 +20,7 @@ final case class WormsNode(
     name: String,
     rank: String,
     aphiaId: Long,
+    acceptedAphiaId: Long,
     alternateNames: Seq[String],
     children: Seq[WormsNode]
 ):
@@ -58,6 +60,7 @@ final case class WormsNode(
     name = this.name,
     rank = this.rank,
     aphiaId = this.aphiaId,
+    acceptedAphiaId = this.acceptedAphiaId,
     alternateNames = this.alternateNames
   )
 
@@ -87,6 +90,7 @@ object WormsNodeBuilder:
       name,
       node.concept.rank,
       node.concept.id,
+      node.concept.acceptedId,
       alternateNames,
       node.children.map(from).toSeq
     )
@@ -95,5 +99,6 @@ final case class SimpleWormsNode(
     name: String,
     rank: String,
     aphiaId: Long,
+    acceptedAphiaId: Long,
     alternateNames: Seq[String]
 )

src/main/scala/org/fathomnet/worms/api/NameEndpoints.scala

@@ -11,7 +11,7 @@ import sttp.tapir._
 import sttp.tapir.json.circe._
 import sttp.tapir.generic.auto._
 import io.circe.generic.auto._
-import org.fathomnet.worms.{Data, NotFound, Page, ServerError, State}
+import org.fathomnet.worms.{Data, Names, NotFound, Page, ServerError, State}
 import sttp.tapir.server.ServerEndpoint
 import scala.concurrent.Future
 import org.fathomnet.worms.ErrorMsg
@@ -54,6 +54,16 @@ class NameEndpoints(using ec: ExecutionContext) extends Endpoints:
   val namesCountServerEndpoint: ServerEndpoint[Any, Future] =
     namesCountEndpoint.serverLogic(Unit => Future.successful(StateController.countAllNames()))
 
+  val namesByAphiaId: PublicEndpoint[Long, ErrorMsg, Names, Any] = baseEndpoint
+    .get
+    .in("names" / "aphiaid")
+    .in(path[Long]("aphiaid"))
+    .out(jsonBody[Names])
+    .description("Returns the data for a name given its aphiaid.")
+
+  val namesByAphiaIdServerEndpoint: ServerEndpoint[Any, Future] =
+    namesByAphiaId.serverLogic(aphiaid => Future(StateController.findNamesByAphiaId(aphiaid)))
+
   // --/query/startswith/:prefix
   val queryStartswithEndpoint: PublicEndpoint[String, ErrorMsg, List[String], Any] = baseEndpoint
     .get
@@ -142,6 +152,7 @@ class NameEndpoints(using ec: ExecutionContext) extends Endpoints:
   override val all: List[ServerEndpoint[Any, Future]] = List(
     namesServerEndpoint,
     namesCountServerEndpoint,
+    namesByAphiaIdServerEndpoint,
     queryStartswithServerEndpoint,
     queryContainsServerEndpoint,
     descendantsServerEndpoint,

src/main/scala/org/fathomnet/worms/api/SwaggerEndpoints.scala

@@ -19,6 +19,7 @@ case class SwaggerEndpoints(nameEndpoints: NameEndpoints, taxaEndpoints: TaxaEnd
         List(
           nameEndpoints.namesCountEndpoint,
           nameEndpoints.namesEndpoint,
+          nameEndpoints.namesByAphiaId,
           nameEndpoints.ancestorsEndpoint,
           nameEndpoints.childrenEndpoint,
           nameEndpoints.descendantsEndpoint,

src/main/scala/org/fathomnet/worms/etc/circe/CirceCodecs.scala

@@ -14,6 +14,7 @@ import org.fathomnet.worms.util.HexUtil
 import org.fathomnet.worms.{ErrorMsg, SimpleWormsNode, WormsNode}
 import org.fathomnet.worms.Page
 import java.net.URI
+import org.fathomnet.worms.Names
 
 /**
  * JSON codecs for use with Circe. Usage:
@@ -48,6 +49,9 @@ object CirceCodecs:
   given Decoder[SimpleWormsNode] = deriveDecoder
   given Encoder[SimpleWormsNode] = deriveEncoder
 
+  given Decoder[Names] = deriveDecoder
+  given Encoder[Names] = deriveEncoder
+
   given Decoder[ErrorMsg] = deriveDecoder
   given Encoder[ErrorMsg] = deriveEncoder
 

src/main/scala/org/fathomnet/worms/io/WormsConcept.scala

@@ -17,11 +17,12 @@ final case class WormsConceptName(name: String, isPrimary: Boolean = true)
 
 final case class WormsConcept(
     id: Long,
+    acceptedId: Long,
     parentId: Option[Long],
     names: Seq[WormsConceptName],
     rank: String,
     isMarine: Boolean = false,
-    isExtinct: Boolean = false
+    isExtinct: Boolean = false,
 )
 
 object WormsConcept:
@@ -36,7 +37,8 @@ object WormsConcept:
   ): Seq[WormsConcept] =
     val concepts = mutable.Map[Long, WormsConcept]()
     for t <- taxons do
-      val wc = WormsConcept(t.id, t.parentId, Seq(WormsConceptName(t.scientificName)), t.rank)
+      val acceptedId = t.acceptedId.getOrElse(t.id)
+      val wc = WormsConcept(t.id, acceptedId, t.parentId, Seq(WormsConceptName(t.scientificName)), t.rank)
       concepts(t.id) = wc
 
     for v <- vernacularNames do

src/main/scala/org/fathomnet/worms/io/extended/CombineTrees.scala

@@ -38,8 +38,9 @@ object CombineTrees:
     rootNode.copy(children = children)
 
   def incrementAphiaId(node: WormsNode, maxAphiaId: Long): WormsNode =
+    val newAphiaId = node.aphiaId + maxAphiaId
     node.copy(
-      aphiaId = node.aphiaId + maxAphiaId,
+      aphiaId = newAphiaId,
       children = node.children.map(incrementAphiaId(_, maxAphiaId))
     )
 
@@ -55,7 +56,10 @@ object CombineTrees:
    *   A new tree with fake aphiaId's flipped to negative
    */
   def flipFakeAphiaId(node: WormsNode, maxAphiaId: Long): WormsNode =
+    val newAphiaId = if (node.aphiaId > maxAphiaId) -node.aphiaId else node.aphiaId
+    val newAcceptedAphiaId = if (newAphiaId < 0) newAphiaId else node.acceptedAphiaId
     node.copy(
-      aphiaId = if (node.aphiaId > maxAphiaId) -node.aphiaId else node.aphiaId,
+      aphiaId = newAphiaId,
+      acceptedAphiaId = newAcceptedAphiaId,
       children = node.children.map(n => flipFakeAphiaId(n, maxAphiaId))
     )

src/main/scala/org/fathomnet/worms/io/extended/ExtendedLoader.scala

@@ -65,7 +65,8 @@ object ExtendedLoader:
         .map(_.trim)
       val conceptNames =
         WormsConceptName(names.head, true) +: names.tail.map(WormsConceptName(_, false))
-      Some(WormsConcept(cols(0).toInt, parentId, conceptNames.toIndexedSeq, ""))
+      val aphiaId = cols(0).toLong
+      Some(WormsConcept(aphiaId, aphiaId, parentId, conceptNames.toIndexedSeq, ""))
     catch
       case NonFatal(e) =>
         log.atWarn.withCause(e).log(s"Failed to parse row: $row")

src/main/scala/org/fathomnet/worms/io/model.scala

@@ -31,17 +31,20 @@ final case class Taxon(
     taxonID: String,
     parentNameUsageID: Option[String],
     scientificName: String,
-    rank: String
+    rank: String,
+    acceptedNameUsageID: Option[String]
 ):
   val id       = taxonIDToKey(taxonID)
   val parentId = parentNameUsageID.map(taxonIDToKey)
+  val acceptedId = acceptedNameUsageID.map(taxonIDToKey)
 
 object Taxon:
   def from(row: String): Option[Taxon] =
     Try {
       val cols              = row.split("\t")
       val parentNameUsageID = if cols(3).isBlank then None else Some(cols(3))
-      Taxon(cols(0), parentNameUsageID, cols(5), cols(19))
+      val acceptedNameUsageID = if cols(2).isBlank then None else Some(cols(2))
+      Taxon(cols(0), parentNameUsageID, cols(5), cols(19), acceptedNameUsageID)
     }.toOption
 
   def read(file: String): List[Taxon] = readFile(file, Taxon.from)

src/test/scala/org/fathomnet/worms/io/extended/CombinedTreesSuite.scala

@@ -12,13 +12,13 @@ import java.nio.file.Paths
 class CombinedTreesSuite extends munit.FunSuite:
 
   test("incrementAphiaId"):
-    val node = WormsNode("foo", "", 1000, Nil, Nil)
+    val node = WormsNode("foo", "", 1000, 1000, Nil, Nil)
     val n = CombineTrees.incrementAphiaId(node, 1000)
     assertEquals(n.aphiaId, 2000L)
 
   test("add"):
-    val node1 = WormsNode("foo", "", 1000, Nil, Nil)
-    val node2 = WormsNode("bar", "", 2000, Nil, Nil)
+    val node1 = WormsNode("foo", "", 1000, 1000, Nil, Nil)
+    val node2 = WormsNode("bar", "", 2000, 2000, Nil, Nil)
     val n = CombineTrees.add(node1, node2, 1000)
     assertEquals(n.aphiaId, 1000L)
     assertEquals(n.children.size, 1)
@@ -27,7 +27,7 @@ class CombinedTreesSuite extends munit.FunSuite:
 
   test("combine") {
 
-    val root = WormsNode("object", "", 1L, Nil, Nil)
+    val root = WormsNode("object", "", 1L, 1L, Nil, Nil)
 
     val sample1 = getClass.getResource("/extended_tree_sample1.csv").getPath
     val opt1 = ExtendedLoader.load(Paths.get(sample1))

src/universal/conf/application.ini

@@ -1,6 +1,6 @@
 # Setting -X directly (-J is stripped)
 # -J-X
--J-Xmx1g
+-J-Xmx2g
 
 # Add additional jvm parameters
 -Duser.timezone=UTC