bring back neo4jcsv import/export, and also: (#182)

* Schema.getNodePropertyNames(nodeLabel: String)
* Schema.getEdgePropertyName(edgeLabel: String)
* various refactorings

core/src/main/scala/flatgraph/DiffGraphApplier.scala

@@ -551,7 +551,10 @@ private[flatgraph] class DiffGraphApplier(graph: Graph, diff: DiffGraphBuilder)
             newPropertyView(insertionBaseIndex + insertionCounter) = insertion.property
           } catch {
             case _: ArrayStoreException =>
-              throw new UnsupportedOperationException(s"unsupported property type: ${insertion.property.getClass}")
+              val edgeType = graph.schema.getEdgeLabel(nodeKind, edgeKind)
+              throw new UnsupportedOperationException(
+                s"unsupported property type `${insertion.property.getClass}` for edge type `$edgeType`"
+              )
           }
         }
         insertionCounter += 1

core/src/main/scala/flatgraph/Edge.scala

@@ -3,6 +3,10 @@ package flatgraph
 class Edge(val src: GNode, val dst: GNode, val edgeKind: Short, val subSeq: Int, val property: Any) {
   def label: String =
     src.graph.schema.getEdgeLabel(src.nodeKind, edgeKind)
+
+  def propertyName: Option[String] = None // override in subclass to provide a property name
+
+  def propertyMaybe: Option[Any] = Option(property)
 }
 
 object Edge {

core/src/main/scala/flatgraph/Graph.scala

@@ -91,13 +91,13 @@ class Graph(val schema: Schema, val storagePathMaybe: Option[Path] = None) exten
   def nodeCount(label: String): Int =
     livingNodeCountByKind(schema.getNodeKindByLabel(label))
 
-  def nodeCount(): Int =
+  def nodeCount: Int =
     livingNodeCountByKind.sum
 
   def allEdges: Iterator[Edge] =
     allNodes.flatMap(Accessors.getEdgesOut)
 
-  def edgeCount(): Int =
+  def edgeCount: Int =
     allEdges.size
 
   /** Lookup nodes with a given label and property value via index. N.b. currently only supported for String properties. Context:
@@ -161,7 +161,7 @@ class Graph(val schema: Schema, val storagePathMaybe: Option[Path] = None) exten
   }
 
   override def toString(): String =
-    s"Graph[${nodeCount()} nodes]"
+    s"Graph[$nodeCount nodes]"
 
   def nodeCountByLabel: Map[String, Int] = {
     schema.nodeKinds

core/src/main/scala/flatgraph/Schema.scala

@@ -111,12 +111,14 @@ abstract class Schema {
   def getNodeKindByLabelMaybe(label: String): Option[Int] = {
     Option(getNodeKindByLabel(label)).filterNot(_ == UndefinedKind)
   }
+  def getNodePropertyNames(nodeLabel: String): Set[String]
 
   // So, the issue here is: We have a couple of pseudo-properties that can only exist at a single node kind
   // (theoretically same for edges). We want to allow our data-layout to alias these properties. This means that multiple
   // properties share the same propertyKind / slot. Hence, to go back from kind -> label, we also need to know the node-kind.
   def getEdgeLabel(nodeKind: Int, edgeKind: Int): String
   def getEdgeKindByLabel(label: String): Int
+  def getEdgePropertyName(label: String): Option[String]
 
   def getPropertyLabel(nodeKind: Int, propertyKind: Int): String
   def getPropertyKindByName(label: String): Int
@@ -149,15 +151,16 @@ abstract class Schema {
 
 class FreeSchema(
   nodeLabels: Array[String],
-  propertyLabels: Array[String],
+  propertyLabels: Array[String], // important: array order corresponds to `nodePropertyPrototypes` order!
   nodePropertyPrototypes: Array[AnyRef],
+  propertyNamesByNodeLabel: Map[String, Set[String]],
   edgeLabels: Array[String],
   edgePropertyPrototypes: Array[AnyRef],
   formalQuantities: Array[FormalQtyType.FormalQuantity] = null
 ) extends Schema {
-  val nodeMap = nodeLabels.zipWithIndex.toMap
-  val propMap = propertyLabels.zipWithIndex.toMap
-  val edgeMap = edgeLabels.zipWithIndex.toMap
+  private val nodeMap = nodeLabels.zipWithIndex.toMap
+  private val propMap = propertyLabels.zipWithIndex.toMap
+  private val edgeMap = edgeLabels.zipWithIndex.toMap
 
   val edgePropertyTypes: Array[FormalQtyType.FormalType] = edgePropertyPrototypes.map(fromPrototype)
   val nodePropertyTypes: Array[FormalQtyType.FormalType] = nodePropertyPrototypes.map(fromPrototype)
@@ -181,10 +184,12 @@ class FreeSchema(
   override def getNodeKindByLabel(label: String): Int                     = nodeMap.getOrElse(label, Schema.UndefinedKind)
   override def getEdgeLabel(nodeKind: Int, edgeKind: Int): String         = edgeLabels(edgeKind)
   override def getEdgeKindByLabel(label: String): Int                     = edgeMap.getOrElse(label, Schema.UndefinedKind)
+  override def getEdgePropertyName(label: String): Option[String]         = None
   override def getPropertyLabel(nodeKind: Int, propertyKind: Int): String = propertyLabels(propertyKind)
   override def getPropertyKindByName(label: String): Int                  = propMap.getOrElse(label, Schema.UndefinedKind)
-  override def getNumberOfPropertyKinds: Int                              = propertyLabels.length
-  override def makeNode(graph: Graph, nodeKind: Short, seq: Int): GNode   = new GNode(graph, nodeKind, seq)
+  override def getNodePropertyNames(nodeLabel: String): Set[String]     = propertyNamesByNodeLabel.getOrElse(nodeLabel, default = Set.empty)
+  override def getNumberOfPropertyKinds: Int                            = propertyLabels.length
+  override def makeNode(graph: Graph, nodeKind: Short, seq: Int): GNode = new GNode(graph, nodeKind, seq)
   override def makeEdge(src: GNode, dst: GNode, edgeKind: Short, subSeq: Int, property: Any): Edge =
     new Edge(src, dst, edgeKind, subSeq, property)
 

core/src/main/scala/flatgraph/storage/Deserialization.scala

@@ -99,9 +99,14 @@ object Deserialization {
   }
 
   private def freeSchemaFromManifest(manifest: Manifest.GraphItem): FreeSchema = {
-    val nodeLabels    = manifest.nodes.map { n => n.nodeLabel }
-    val nodePropNames = mutable.LinkedHashMap[String, AnyRef]()
+    val nodeLabels               = manifest.nodes.map { n => n.nodeLabel }
+    val nodePropNames            = mutable.LinkedHashMap.empty[String, AnyRef]
+    val propertyNamesByNodeLabel = mutable.LinkedHashMap.empty[String, Set[String]]
     for (prop <- manifest.properties) {
+      propertyNamesByNodeLabel.updateWith(prop.nodeLabel) {
+        case None             => Some(Set(prop.propertyLabel))
+        case Some(oldEntries) => Some(oldEntries + prop.propertyLabel)
+      }
       nodePropNames(prop.propertyLabel) = protoFromOutline(prop.property)
     }
     val propertyLabels         = nodePropNames.keysIterator.toArray
@@ -117,7 +122,7 @@ object Deserialization {
     val edgeLabels             = edgePropNames.keysIterator.toArray
     val edgePropertyPrototypes = edgePropNames.valuesIterator.toArray
 
-    new FreeSchema(nodeLabels, propertyLabels, nodePropertyPrototypes, edgeLabels, edgePropertyPrototypes)
+    new FreeSchema(nodeLabels, propertyLabels, nodePropertyPrototypes, propertyNamesByNodeLabel.toMap, edgeLabels, edgePropertyPrototypes)
   }
 
   private def protoFromOutline(outline: OutlineStorage): AnyRef = {

core/src/main/scala/flatgraph/traversal/Language.scala

@@ -470,6 +470,18 @@ class NodeMethods(node: GNode) extends AnyVal {
     }
   }
 
+  def propertiesMap: java.util.Map[String, AnyRef] = {
+    val ret    = new java.util.HashMap[String, AnyRef]()
+    val schema = node.graph.schema
+    schema.propertyKinds.foreach { propertyKind =>
+      Accessors.getNodePropertyOptionCompat(node, propertyKind).foreach { value =>
+        val key = schema.getPropertyLabel(node.nodeKind, propertyKind)
+        ret.put(key, value)
+      }
+    }
+    ret
+  }
+
   private def edgeKind(edgeLabel: String): Int =
     node.graph.schema.getEdgeKindByLabel(edgeLabel)
 }

core/src/test/scala/flatgraph/GraphTests.scala

@@ -738,6 +738,7 @@ class GraphTests extends AnyWordSpec with Matchers {
       nodeLabels = Array(Node0Label),
       edgeLabels = Array.empty,
       propertyLabels = Array(propertySingle, propertyOptional, propertyMulti).map(_.name),
+      propertyNamesByNodeLabel = Map.empty,
       edgePropertyPrototypes = new Array[AnyRef](0),
       nodePropertyPrototypes = Array(new Array[String](0), new Array[String](0), new Array[String](0)),
       formalQuantities = Array(FormalQtyType.QtyOne, null, FormalQtyType.QtyOption, null, FormalQtyType.QtyMulti)

core/src/test/scala/flatgraph/TestHelpers.scala

@@ -44,6 +44,7 @@ object TestSchema {
       nodeLabels = Range(0, nodeKinds).map { id => s"V${id}" }.toArray,
       edgeLabels = Range(0, edgeKinds).map { id => s"${id}" }.toArray,
       propertyLabels = Range(0, properties).map { id => s"${id}" }.toArray,
+      propertyNamesByNodeLabel = Map.empty,
       edgePropertyPrototypes = if (edgePropertyPrototypes != null) edgePropertyPrototypes else new Array[AnyRef](edgeKinds),
       nodePropertyPrototypes = if (nodePropertyPrototypes != null) nodePropertyPrototypes else new Array[AnyRef](properties),
       formalQtys

core/src/test/scala/flatgraph/traversal/testdomains/simple/ExampleGraphSetup.scala

@@ -92,6 +92,7 @@ object SimpleDomain {
       nodeLabels = Array(Thing.Label),
       edgeLabels = edgeLabels,
       propertyLabels = Array(ExampleGraphSetup.Properties.Name.name),
+      propertyNamesByNodeLabel = Map.empty,
       edgePropertyPrototypes = new Array(edgeLabels.length),
       nodePropertyPrototypes = Array(Array.empty[String])
     )

domain-classes-generator/src/main/scala/flatgraph/codegen/DomainClassesGenerator.scala

@@ -51,7 +51,7 @@ class DomainClassesGenerator(schema: Schema) {
 
     val propertyKindByProperty = relevantProperties.zipWithIndex.toMap
     val edgeTypes              = schema.edgeTypes.sortBy(_.name).toArray
-    val edgeIdByType           = edgeTypes.zipWithIndex.toMap
+    val edgeKindByType         = edgeTypes.zipWithIndex.toMap
 
     val newPropertiesByNodeType: Map[AbstractNodeType, Set[Property[?]]] =
       schema.allNodeTypes.map { nodeType =>
@@ -82,8 +82,8 @@ class DomainClassesGenerator(schema: Schema) {
     // format: off
     val edgeAccess = edgeTypes.map { et =>
       s"""
-       |final def _${camelCase(et.name)}Out: Iterator[StoredNode] = flatgraph.Accessors.getNeighborsOut(this.graph, this.nodeKind, this.seq, ${edgeIdByType(et)}).asInstanceOf[Iterator[StoredNode]]
-       |final def _${camelCase(et.name)}In: Iterator[StoredNode] = flatgraph.Accessors.getNeighborsIn(this.graph, this.nodeKind, this.seq, ${edgeIdByType(et)}).asInstanceOf[Iterator[StoredNode]]
+       |final def _${camelCase(et.name)}Out: Iterator[StoredNode] = flatgraph.Accessors.getNeighborsOut(this.graph, this.nodeKind, this.seq, ${edgeKindByType(et)}).asInstanceOf[Iterator[StoredNode]]
+       |final def _${camelCase(et.name)}In: Iterator[StoredNode] = flatgraph.Accessors.getNeighborsIn(this.graph, this.nodeKind, this.seq, ${edgeKindByType(et)}).asInstanceOf[Iterator[StoredNode]]
        |""".stripMargin
     }.mkString("\n")
     // format: on
@@ -214,27 +214,39 @@ class DomainClassesGenerator(schema: Schema) {
         if (edgeType.properties.length > 1) throw new RuntimeException("we only support zero or one edge properties")
 
         // format: off
-        val propertyAccessor = if (edgeType.properties.length == 1) {
-          val p = edgeType.properties.head
-          p.cardinality match {
-            case _: Cardinality.One[?] =>
-              s"""{
-                 |  def ${camelCase(p.name)}: ${unpackTypeUnboxed(p.valueType, true )} = this.property.asInstanceOf[${unpackTypeUnboxed(p.valueType, true)}]
-                 |}""".stripMargin
-            case Cardinality.ZeroOrOne =>
-              s"""{
-                 |  def ${camelCase(p.name)}: Option[${unpackTypeUnboxed(p.valueType, true)}] = Option(this.property.asInstanceOf[${unpackTypeBoxed(p.valueType, true)}])
-                 |}""".stripMargin
-            case Cardinality.List => throw new RuntimeException("edge properties are only supported with cardinality one or optional")
-          }
-        } else ""
+        val propertyAccessorMaybe: Option[String] = edgeType.properties match {
+          case Nil => None
+          case Seq(p) =>
+            p.cardinality match {
+              case _: Cardinality.One[?] =>
+                Some(s"""def ${camelCase(p.name)}: ${unpackTypeUnboxed(p.valueType, true )} =
+                   |  this.property.asInstanceOf[${unpackTypeUnboxed(p.valueType, true)}]""".stripMargin)
+              case Cardinality.ZeroOrOne =>
+                Some(s"""def ${camelCase(p.name)}: Option[${unpackTypeUnboxed(p.valueType, true)}] =
+                   |  Option(this.property.asInstanceOf[${unpackTypeBoxed(p.valueType, true)}])""".stripMargin)
+              case Cardinality.List =>
+                throw new RuntimeException("edge properties are only supported with cardinality one or optional")
+            }
+          case multipleProperties =>
+            throw new RuntimeException(s"only one edge property supported - ${edgeType.name} defines ${multipleProperties.size}")
+        }
+
+        val propertyNameImplForObject = edgeType.properties.headOption.map { property=>
+          s"""val propertyName: Option[String] = Some("${property.name}")"""
+        }
+        val propertyNameImplForClass = propertyNameImplForObject.map { _ =>
+          s"override def propertyName: Option[String] = ${edgeType.className}.propertyName"
+        }
 
         s"""object ${edgeType.className} {
            |  val Label = "${edgeType.name}"
+           |  ${propertyNameImplForObject.getOrElse("")}
            |}
            |
            |class ${edgeType.className}(src_4762: flatgraph.GNode, dst_4762: flatgraph.GNode, subSeq_4862: Int, property_4862: Any)
-           |  extends flatgraph.Edge(src_4762, dst_4762, ${edgeKindByEdgeType(edgeType)}.toShort, subSeq_4862, property_4862) $propertyAccessor
+           |  extends flatgraph.Edge(src_4762, dst_4762, ${edgeKindByEdgeType(edgeType)}.toShort, subSeq_4862, property_4862) {
+           |  ${propertyNameImplForClass.getOrElse("")}
+           |}
            |""".stripMargin
         // format: on
       }
@@ -605,30 +617,54 @@ class DomainClassesGenerator(schema: Schema) {
         sourceLines.result()
       }
 
+      val nodePropertyNameCases = for {
+        nodeType <- nodeTypes
+        propertyNames = nodeType.properties.map(p => s""""${p.name}"""").mkString(", ")
+      } yield s"""case "${nodeType.name}" => Set($propertyNames)"""
+
+      val edgePropertyNameCases = for {
+        edgeType <- edgeTypes
+        property <- edgeType.properties.headOption
+      } yield s"""case "${edgeType.name}" => Some("${property.name}")"""
+
+      // format: off
       s"""package $basePackage
+         |
          |import $basePackage.nodes
          |import $basePackage.edges
          |import flatgraph.FormalQtyType
          |
          |object GraphSchema extends flatgraph.Schema {
          |  private val nodeLabels = IndexedSeq($nodeLabelsSrc)
          |  val nodeKindByLabel = nodeLabels.zipWithIndex.toMap
-         |  val edgeLabels = Array(${edgeTypes.map { e => s""""${e.name}"""" }.mkString(",  ")})
-         |  val edgeIdByLabel = edgeLabels.zipWithIndex.toMap
+         |  val edgeLabels = Array(${edgeTypes.map { e => s""""${e.name}"""" }.mkString(", ")})
+         |  val edgeKindByLabel = edgeLabels.zipWithIndex.toMap
          |  val edgePropertyAllocators: Array[Int => Array[?]] = Array($edgePropertyAllocatorsSrc)
          |  val nodeFactories: Array[(flatgraph.Graph, Int) => nodes.StoredNode] = Array($nodeFactoriesSrc)
          |  val edgeFactories: Array[(flatgraph.GNode, flatgraph.GNode, Int, Any) => flatgraph.Edge] = Array($edgeFactoriesSrc)
          |  val nodePropertyAllocators: Array[Int => Array[?]] = Array($nodePropertyAllocatorsSrc)
          |  val normalNodePropertyNames = Array(${relevantProperties.map { p => s""""${p.name}"""" }.mkString(", ")})
          |  val nodePropertyByLabel = normalNodePropertyNames.zipWithIndex.toMap$nodePropertyByLabelSrc
-         |  val nodePropertyDescriptors: Array[FormalQtyType.FormalQuantity | FormalQtyType.FormalType] = ${nodePropertyDescriptorsSource
-          .mkString("\n")}
+         |  val nodePropertyDescriptors: Array[FormalQtyType.FormalQuantity | FormalQtyType.FormalType] = ${nodePropertyDescriptorsSource.mkString("\n")}
          |  override def getNumberOfNodeKinds: Int = ${nodeTypes.length}
          |  override def getNumberOfEdgeKinds: Int = ${edgeTypes.length}
          |  override def getNodeLabel(nodeKind: Int): String = nodeLabels(nodeKind)
          |  override def getNodeKindByLabel(label: String): Int = nodeKindByLabel.getOrElse(label, flatgraph.Schema.UndefinedKind)
          |  override def getEdgeLabel(nodeKind: Int, edgeKind: Int): String = edgeLabels(edgeKind)
-         |  override def getEdgeKindByLabel(label: String): Int = edgeIdByLabel.getOrElse(label, flatgraph.Schema.UndefinedKind)
+         |  override def getEdgeKindByLabel(label: String): Int = edgeKindByLabel.getOrElse(label, flatgraph.Schema.UndefinedKind)
+         |  override def getNodePropertyNames(nodeLabel: String): Set[String] = {
+         |    nodeLabel match {
+         |      ${nodePropertyNameCases.mkString("\n")}
+         |      case _ => Set.empty
+         |    }
+         |  }
+         |  override def getEdgePropertyName(label: String): Option[String] = {
+         |    label match {
+         |      ${edgePropertyNameCases.mkString("\n")}
+         |      case _ => None
+         |    }
+         |  }
+         |
          |  override def getPropertyLabel(nodeKind: Int, propertyKind: Int): String = {
          |    if(propertyKind < ${relevantProperties.length}) normalNodePropertyNames(propertyKind)
          |    $containedNodesAsPropertyCases
@@ -643,6 +679,7 @@ class DomainClassesGenerator(schema: Schema) {
          |  override def getNodePropertyFormalType(nodeKind: Int, propertyKind: Int): FormalQtyType.FormalType = nodePropertyDescriptors(propertyOffsetArrayIndex(nodeKind, propertyKind)).asInstanceOf[FormalQtyType.FormalType]
          |  override def getNodePropertyFormalQuantity(nodeKind: Int, propertyKind: Int): FormalQtyType.FormalQuantity = nodePropertyDescriptors(1 + propertyOffsetArrayIndex(nodeKind, propertyKind)).asInstanceOf[FormalQtyType.FormalQuantity]
          |}""".stripMargin
+      // format: on
     }
     os.write(outputDir0 / "GraphSchema.scala", schemaFile)
 

formats-tests/src/test/resources/neo4jcsv/edges_connected_to_cypher.csv

@@ -0,0 +1,4 @@
+LOAD CSV FROM 'file:/edges_connected_to_data.csv' AS line
+MATCH (a), (b)
+WHERE a.id = toInteger(line[0]) AND b.id = toInteger(line[1])
+CREATE (a)-[r:connected_to {string_mandatory: line[3]}]->(b);

formats-tests/src/test/resources/neo4jcsv/edges_connected_to_data.csv

@@ -0,0 +1 @@
+0,1,connected_to,edge property

formats-tests/src/test/resources/neo4jcsv/edges_connected_to_header.csv

@@ -0,0 +1 @@
+:START_ID,:END_ID,:TYPE,string_mandatory:string

formats-tests/src/test/resources/neo4jcsv/nodes_node_a_cypher.csv

@@ -0,0 +1,10 @@
+LOAD CSV FROM 'file:/nodes_node_a_data.csv' AS line
+CREATE (:node_a {
+id: toInteger(line[0]),
+int_list: toIntegerList(split(line[2], ";")),
+int_mandatory: toInteger(line[3]),
+int_optional: toInteger(line[4]),
+string_list: toStringList(split(line[5], ";")),
+string_mandatory: line[6],
+string_optional: line[7]
+});

formats-tests/src/test/resources/neo4jcsv/nodes_node_a_data.csv

@@ -0,0 +1,2 @@
+0,node_a,,42,,node 3 c1;node 3 c2,node 2 a,node 2 b
+1,node_a,10;11;12,1,2,,<empty>,

formats-tests/src/test/resources/neo4jcsv/nodes_node_a_header.csv

@@ -0,0 +1 @@
+:ID,:LABEL,int_list:int[],int_mandatory:int,int_optional:int,string_list:string[],string_mandatory:string,string_optional:string

formats-tests/src/test/scala/flatgraph/TestDomainSimple.scala

@@ -26,6 +26,7 @@ object TestDomainSimple {
       PropertyNames.ContainedTestNodeProperty
     ),
     nodePropertyPrototypes = Array(Array.empty[String], Array.empty[String], Array.emptyIntArray, Array.emptyIntArray, Array.empty[GNode]),
+    propertyNamesByNodeLabel = Map.empty,
     edgePropertyPrototypes = Array(Array.emptyLongArray)
   )