log node/edge counts when writing graphs (#320)

* log node/edge counts when writing graphs

* collect counts during serialization to avoid walking graph twice

* also log counts during deserialization

Author: maltek

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

@@ -157,7 +157,8 @@ class Graph(val schema: Schema, val storagePathMaybe: Option[Path] = None) exten
 
     storagePathMaybe.foreach { storagePath =>
       logger.info(s"closing graph: writing to storage at `$storagePath`")
-      Serialization.writeGraph(this, storagePath)
+      val (nodes, edges, props) = Serialization.writeGraph(this, storagePath)
+      logger.debug(s"closed graph: wrote $nodes nodes with $edges edges and $props properties")
     }
   }
 

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

@@ -4,6 +4,7 @@ import flatgraph.{AccessHelpers, FreeSchema, GNode, Graph, Schema}
 import flatgraph.Edge.Direction
 import flatgraph.misc.Misc
 import flatgraph.storage.Manifest.{GraphItem, OutlineStorage}
+import org.slf4j.LoggerFactory
 
 import java.nio.channels.FileChannel
 import java.nio.charset.StandardCharsets
@@ -14,6 +15,7 @@ import scala.collection.mutable
 import java.util.concurrent
 
 object Deserialization {
+  private val logger = LoggerFactory.getLogger(classOf[Deserialization.type])
 
   def readGraph(
     storagePath: Path,
@@ -34,8 +36,11 @@ object Deserialization {
     try {
       // fixme: Use convenience methods from schema to translate string->id. Fix after we get strict schema checking.
       val manifest = GraphItem.read(readManifest(fileChannel))
-      val pool     = submitJob { readPool(manifest, fileChannel, zstdCtx) }
-      val schema   = schemaMaybe.getOrElse(freeSchemaFromManifest(manifest))
+      logger.debug(
+        s"loading graph with ${manifest.nodes.length} nodes, ${manifest.edges.length} edges, ${manifest.properties.length} properties"
+      )
+      val pool   = submitJob { readPool(manifest, fileChannel, zstdCtx) }
+      val schema = schemaMaybe.getOrElse(freeSchemaFromManifest(manifest))
       val storagePathMaybe =
         if (persistOnClose) Option(storagePath)
         else None

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

@@ -220,7 +220,7 @@ class WriterContext(val fileChannel: FileChannel, val executor: concurrent.Execu
 
 object Serialization {
 
-  def writeGraph(g: Graph, storagePath: Path, requestedExecutor: Option[concurrent.ExecutorService] = None): Unit = {
+  def writeGraph(g: Graph, storagePath: Path, requestedExecutor: Option[concurrent.ExecutorService] = None): (Int, Int, Int) = {
 
     // ensure parent directory exists
     Option(storagePath.getParent) match {
@@ -243,7 +243,7 @@ object Serialization {
     }
   }
 
-  private def innerWriteGraph(g: Graph, writer: WriterContext): Unit = {
+  private def innerWriteGraph(g: Graph, writer: WriterContext): (Int, Int, Int) = {
     val nodes      = mutable.ArrayBuffer.empty[NodeItem]
     val edges      = mutable.ArrayBuffer.empty[EdgeItem]
     val properties = mutable.ArrayBuffer.empty[PropertyItem]
@@ -288,9 +288,11 @@ object Serialization {
       }
     }
 
+    val counts   = (nodes.size, edges.size, properties.size)
     val manifest = new GraphItem(nodes.toArray, edges.toArray, properties.toArray)
     writer.finish(manifest)
 
+    counts
   }
 
   private[flatgraph] def write(bytes: Array[Byte], res: OutlineStorage, filePtr: AtomicLong, fileChannel: FileChannel): OutlineStorage = {