|
| 1 | +package hmda.loader.lar |
| 2 | + |
| 3 | +import java.io.File |
| 4 | +import java.time.Instant |
| 5 | + |
| 6 | +import akka.pattern.ask |
| 7 | +import akka.actor.{ ActorPath, ActorRef, ActorSystem } |
| 8 | +import akka.cluster.client.{ ClusterClient, ClusterClientSettings } |
| 9 | +import akka.stream.{ ActorMaterializer, IOResult } |
| 10 | +import akka.stream.scaladsl.{ FileIO, Sink, Source } |
| 11 | +import akka.util.{ ByteString, Timeout } |
| 12 | +import hmda.api.util.FlowUtils |
| 13 | +import hmda.model.fi.{ Created, Filing, Submission } |
| 14 | +import hmda.persistence.HmdaSupervisor.{ FindFilings, FindHmdaFiling, FindProcessingActor, FindSubmissions } |
| 15 | +import hmda.persistence.institutions.FilingPersistence.CreateFiling |
| 16 | +import hmda.persistence.institutions.SubmissionPersistence.CreateSubmission |
| 17 | +import hmda.persistence.institutions.{ FilingPersistence, SubmissionPersistence } |
| 18 | +import hmda.persistence.processing.HmdaRawFile.AddLine |
| 19 | +import hmda.persistence.processing.ProcessingMessages.{ CompleteUpload, Persisted, StartUpload } |
| 20 | +import hmda.persistence.processing.SubmissionManager |
| 21 | +import hmda.persistence.messages.CommonMessages._ |
| 22 | +import hmda.persistence.processing.SubmissionManager.AddFileName |
| 23 | +import hmda.query.HmdaQuerySupervisor.FindHmdaFilingView |
| 24 | +import org.slf4j.LoggerFactory |
| 25 | + |
| 26 | +import scala.concurrent.duration._ |
| 27 | +import scala.concurrent.{ ExecutionContext, Future } |
| 28 | +import scala.util.{ Failure, Success } |
| 29 | + |
| 30 | +object HmdaBatchLarLoader extends FlowUtils { |
| 31 | + |
| 32 | + override implicit val system: ActorSystem = ActorSystem("hmda-cluster-client") |
| 33 | + override implicit val materializer: ActorMaterializer = ActorMaterializer() |
| 34 | + override implicit val ec: ExecutionContext = system.dispatcher |
| 35 | + |
| 36 | + val hmdaClusterName = config.getString("hmda.clusterName") |
| 37 | + val hmdaClusterIP = config.getString("hmda.lar.host") |
| 38 | + val hmdaClusterPort = config.getInt("hmda.lar.port") |
| 39 | + val actorTimeout = config.getInt("hmda.actorTimeout") |
| 40 | + val flowParallelism = config.getInt("hmda.lar.parallelism") |
| 41 | + |
| 42 | + implicit val timeout = Timeout(actorTimeout.seconds) |
| 43 | + |
| 44 | + val log = LoggerFactory.getLogger("hmda-lar-loader") |
| 45 | + |
| 46 | + val initialContacts = Set( |
| 47 | + ActorPath.fromString(s"akka.tcp://$hmdaClusterName@$hmdaClusterIP:$hmdaClusterPort/system/receptionist") |
| 48 | + ) |
| 49 | + |
| 50 | + val settings = ClusterClientSettings(system) |
| 51 | + .withInitialContacts(initialContacts) |
| 52 | + |
| 53 | + val clusterClient = system.actorOf(ClusterClient.props(settings), "hmda-lar-loader") |
| 54 | + |
| 55 | + def main(args: Array[String]): Unit = { |
| 56 | + |
| 57 | + if (args.length < 2) { |
| 58 | + exitSys(log, "Please provide a directory containing files with LAR data and period to process", 1) |
| 59 | + } |
| 60 | + |
| 61 | + val path = new File(args(0)) |
| 62 | + if (!path.isDirectory) { |
| 63 | + exitSys(log, "Argument must be the full path to a directory containing files with LAR data", 2) |
| 64 | + } |
| 65 | + |
| 66 | + val period = args(1) |
| 67 | + |
| 68 | + val fileList = path.listFiles().toSet |
| 69 | + val fileNames = fileList.map(file => file.getName).filter(name => name.endsWith(".txt")) |
| 70 | + |
| 71 | + fileNames.foreach(fileName => processLars(new File(s"$path/$fileName"), s"$fileName", s"${fileName.substring(0, fileName.indexOf("."))}", period)) |
| 72 | + |
| 73 | + } |
| 74 | + |
| 75 | + private def processLars(file: File, fileName: String, institutionId: String, period: String) = { |
| 76 | + val uploadTimestamp = Instant.now.toEpochMilli |
| 77 | + val source = FileIO.fromPath(file.toPath) |
| 78 | + |
| 79 | + val fSubmissionsActor = (clusterClient ? ClusterClient |
| 80 | + .Send( |
| 81 | + "/user/supervisor/singleton", |
| 82 | + FindSubmissions(SubmissionPersistence.name, institutionId, period), |
| 83 | + localAffinity = true |
| 84 | + )).mapTo[ActorRef] |
| 85 | + |
| 86 | + (clusterClient ? ClusterClient.Send("/user/supervisor/singleton", FindHmdaFiling(period), localAffinity = true)).mapTo[ActorRef] |
| 87 | + (clusterClient ? ClusterClient.Send("/user/query-supervisor", FindHmdaFilingView(period), localAffinity = true)).mapTo[ActorRef] |
| 88 | + |
| 89 | + val fUploadSubmission = for { |
| 90 | + s <- fSubmissionsActor |
| 91 | + fSubmission <- (s ? CreateSubmission).mapTo[Option[Submission]] |
| 92 | + submission = fSubmission.getOrElse(Submission()) |
| 93 | + } yield (submission, submission.status == Created) |
| 94 | + |
| 95 | + fUploadSubmission.onComplete { |
| 96 | + case Success((submission, true)) => |
| 97 | + val findSubmissionManager = FindProcessingActor(SubmissionManager.name, submission.id) |
| 98 | + val findFilingPersistence = FindFilings(FilingPersistence.name, institutionId) |
| 99 | + val fProcessingActor = (clusterClient ? ClusterClient.Send("/user/supervisor/singleton", findSubmissionManager, localAffinity = true)).mapTo[ActorRef] |
| 100 | + val fFilingPersistence = (clusterClient ? ClusterClient.Send("/user/supervisor/singleton", findFilingPersistence, localAffinity = true)).mapTo[ActorRef] |
| 101 | + |
| 102 | + for { |
| 103 | + f <- fFilingPersistence |
| 104 | + p <- fProcessingActor |
| 105 | + _ <- (f ? CreateFiling(Filing(period, institutionId))).mapTo[Option[Filing]] |
| 106 | + } yield { |
| 107 | + uploadData(p, uploadTimestamp, fileName, submission, source) |
| 108 | + } |
| 109 | + |
| 110 | + case Success((_, false)) => |
| 111 | + log.error("submission not available for upload") |
| 112 | + sys.exit(0) |
| 113 | + |
| 114 | + case Failure(error) => |
| 115 | + log.error(error.getLocalizedMessage) |
| 116 | + sys.exit(0) |
| 117 | + |
| 118 | + } |
| 119 | + } |
| 120 | + |
| 121 | + private def uploadData(processingActor: ActorRef, uploadTimestamp: Long, fileName: String, submission: Submission, source: Source[ByteString, Future[IOResult]]): Unit = { |
| 122 | + processingActor ! AddFileName(fileName) |
| 123 | + processingActor ! StartUpload |
| 124 | + val uploadedF = source |
| 125 | + .via(framing) |
| 126 | + .map(_.utf8String) |
| 127 | + .mapAsync(parallelism = flowParallelism)(line => (processingActor ? AddLine(uploadTimestamp, line)).mapTo[Persisted.type]) |
| 128 | + .runWith(Sink.ignore) |
| 129 | + |
| 130 | + uploadedF.onComplete { |
| 131 | + case Success(_) => |
| 132 | + processingActor ! CompleteUpload |
| 133 | + |
| 134 | + case Failure(error) => |
| 135 | + processingActor ! Shutdown |
| 136 | + log.error(error.getLocalizedMessage) |
| 137 | + } |
| 138 | + } |
| 139 | + |
| 140 | +} |
0 commit comments