├── project ├── build.properties └── plugins.sbt ├── .gitignore ├── src └── main │ ├── scala │ ├── com │ │ └── bayakala │ │ │ └── funda │ │ │ ├── fda │ │ │ └── package.scala │ │ │ ├── fdapipes │ │ │ ├── FDAJoints.scala │ │ │ ├── FDAValves.scala │ │ │ ├── FDANodes.scala │ │ │ └── FDAPipes.scala │ │ │ ├── fdasources │ │ │ ├── FDADataView.scala │ │ │ └── FDADataStream.scala │ │ │ ├── fdapars │ │ │ └── FDAPars.scala │ │ │ ├── fdarows │ │ │ └── FDARowOperator.scala │ │ │ └── package.scala │ ├── examples │ │ ├── ExceptionsAndFinalizers.scala │ │ ├── StrongTypedSource.scala │ │ ├── ParallelLoading.scala │ │ ├── UserDefinedTask.scala │ │ ├── ParallelTasks.scala │ │ └── ParallelExecution.scala │ └── samples │ │ └── SlickModels.scala │ └── resources │ ├── application.conf │ └── logback.xml ├── license.txt └── README.md /project/build.properties: -------------------------------------------------------------------------------- 1 | sbt.version = 0.13.8 2 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | target/ 2 | project/target/ 3 | .idea 4 | .DS_Store 5 | -------------------------------------------------------------------------------- /project/plugins.sbt: -------------------------------------------------------------------------------- 1 | addSbtPlugin("me.lessis" % "bintray-sbt" % "0.3.0") 2 | logLevel := Level.Warn 3 | -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fda/package.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda 2 | import com.bayakala.funda.fdapipes._ 3 | import com.bayakala.funda.fdarows._ 4 | import com.bayakala.funda.fdasources._ 5 | import com.bayakala.funda.fdapars._ 6 | 7 | package object api extends FDARowOperator 8 | with FDAPipes 9 | with FDADataView 10 | with FDADataStream 11 | with FDAPars 12 | 13 | 14 | -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdapipes/FDAJoints.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdapipes 2 | 3 | import com.bayakala.funda.FDAPipeJoint 4 | import fs2._ 5 | 6 | /** 7 | * methods to output row to a Pull object 8 | */ 9 | object FDAJoints { //数据发送方法 10 | /** 11 | * send row to Pull 12 | * @param row row to be sent 13 | * @tparam ROW type of row 14 | * @return new state of Pull 15 | */ 16 | def fda_pushRow[ROW](row: ROW): FDAPipeJoint[ROW] = Pull.output1(row) 17 | 18 | /** 19 | * send list of rows to Pull 20 | * @param rows list of rows to be sent 21 | * @tparam ROW type of target row 22 | * @return new state of Pull 23 | */ 24 | def fda_pushRows[ROW](rows: List[ROW]): FDAPipeJoint[ROW] = Pull.output(Chunk.seq(rows)) 25 | 26 | /** 27 | * end output to Pull 28 | * @return end of Pull 29 | */ 30 | def fda_halt = Pull.done 31 | } 32 | -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdapipes/FDAValves.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdapipes 2 | 3 | /** 4 | * methods to send data downstream 5 | * Some(Nil) : skip sending the current row 6 | * Some(List(r1,r2...)): send r1,r2... downstream 7 | * None : halt stream, end of process 8 | * 9 | */ 10 | object FDAValves { //流动控制方法 11 | /** 跳过本行（不向下游发送） 12 | * skip current row by sending a Some(Nil) 13 | * @return Some(Nil) to signify skipping current row 14 | */ 15 | def fda_skip[ROW]: Option[List[ROW]] = Some(List[ROW]()) 16 | 17 | /** 将本行发送至下游连接管道 18 | * send row downstream 19 | * @param row target row 20 | * @tparam ROW type of target row 21 | * @return a single row to be sent downstream 22 | */ 23 | def fda_next[ROW](row: ROW): Option[List[ROW]] = Some(List[ROW](row)) 24 | 25 | /** 26 | * send a list of rows downstream in a chunk 27 | * @param lr list of rows to be sent 28 | * @tparam ROW type of target row 29 | * @return a list of many rows 30 | */ 31 | def fda_next[ROW](lr: List[ROW]): Option[List[ROW]] = Some(lr) 32 | 33 | /** 终止流动 34 | * halt the current stream 35 | * @return a None indicating end of current stream process 36 | */ 37 | def fda_break = None 38 | 39 | } 40 | 41 | -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdasources/FDADataView.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdasources 2 | 3 | import com.bayakala.funda._ 4 | import fs2._ 5 | 6 | /** 7 | * produce a static view source from a Seq[ROW] collection returned from 8 | * running a fda_typedRows function. provide error handling and cleanup 9 | */ 10 | trait FDADataView { 11 | /** 12 | * return a fs2 stream by closing a Pull object 13 | * @param h a Seq[ROW] type collection 14 | * @tparam ROW type of element inside collection 15 | * @return an effectful stream 16 | */ 17 | private def streamSeq[ROW](h: Seq[ROW]): FDAPipeLine[ROW] = 18 | pullSeq(h).close 19 | 20 | /** 21 | * emit a collection of rows one by one through a Pull object. 22 | * could use fold to reduce Seq[ROW] to Stream[Task,ROW] like this: 23 | * foldLeft(Stream[Task,Int]())((s,a) => s ++ Stream.emit(a)) 24 | * but since h could be huge and recursion safety could not be guaranteed 25 | * streaming through Pull object is a much safer option because 26 | * Pull is a free monad 27 | * @param h a Seq[ROW] type collection 28 | * @tparam ROW type of element inside collection 29 | * @return Pull in new state 30 | */ 31 | private def pullSeq[ROW](h: Seq[ROW]): FDAPipeJoint[ROW] = { 32 | val it = h.iterator 33 | def go(it: Iterator[ROW]): Pull[Task, ROW, Unit] = for { 34 | res <- Pull.eval(Task.delay({ if (it.hasNext) Some(it.next()) else None })) 35 | next <- res.fold[Pull[Task, ROW, Unit]](Pull.done)(o => Pull.output1(o) >> go(it)) 36 | } yield next 37 | go(it) 38 | } 39 | 40 | /** 41 | * produce a static view source from a Seq[ROW] collection using famous 'bracket' 42 | * provide facade to error handling and cleanup 43 | * @example {{{ 44 | * val source = fda_staticSource(dataSeq)() 45 | * 46 | * val safeSource = fda_staticSource(dataSeq)( 47 | * println("the end finally!")) 48 | * }}} 49 | * @param acquirer the Seq[ROW] collection 50 | * @param finalizer cleanup callback 51 | * @tparam ROW type of row 52 | * @return a new stream 53 | */ 54 | def fda_staticSource[ROW](acquirer: => Seq[ROW])( 55 | finalizer: => Unit = ()): FDAPipeLine[ROW] = { 56 | val s = Stream.bracket(Task.delay(acquirer))(r => streamSeq(r), r => Task.delay((): Unit)) 57 | s.onFinalize(Task.delay(finalizer)) 58 | } 59 | 60 | } 61 | 62 | /** 63 | * for global imports 64 | */ 65 | object FDADataView extends FDADataView 66 | -------------------------------------------------------------------------------- /src/main/resources/application.conf: -------------------------------------------------------------------------------- 1 | app = { 2 | dbconfig = h2 3 | } 4 | 5 | logger.scala.slick.jdbc.JdbcBackend.statement=DEBUG 6 | 7 | h2 { 8 | driver = "slick.driver.H2Driver$" 9 | db { 10 | url = "jdbc:h2:~/slickdemo;mv_store=false" 11 | driver = "org.h2.Driver" 12 | connectionPool = HikariCP 13 | numThreads = 10 14 | maxConnections = 12 15 | minConnections = 4 16 | keepAliveConnection = true 17 | } 18 | } 19 | 20 | h2db { 21 | url = "jdbc:h2:tcp://localhost/~/slickdemo;mv_store=false" 22 | driver = "org.h2.Driver" 23 | connectionPool = HikariCP 24 | numThreads = 48 25 | maxConnections = 48 26 | minConnections = 12 27 | keepAliveConnection = true 28 | } 29 | 30 | h2mem = { 31 | url = "jdbc:h2:mem:slickdemo" 32 | driver = org.h2.Driver 33 | connectionPool = disabled 34 | keepAliveConnection = true 35 | } 36 | 37 | mysql { 38 | driver = "slick.driver.MySQLDriver$" 39 | db { 40 | url = "jdbc:mysql://localhost/slickdemo" 41 | driver = com.mysql.jdbc.Driver 42 | keepAliveConnection = true 43 | user="root" 44 | password="123" 45 | numThreads=10 46 | maxConnections = 12 47 | minConnections = 4 48 | } 49 | } 50 | 51 | mysqldb = { 52 | dataSourceClass = "com.mysql.jdbc.jdbc2.optional.MysqlDataSource" 53 | properties { 54 | user = "root" 55 | password = "123" 56 | databaseName = "slickdemo" 57 | serverName = "localhost" 58 | } 59 | numThreads = 10 60 | maxConnections = 12 61 | minConnections = 4 62 | } 63 | 64 | postgres { 65 | driver = "slick.driver.PostgresDriver$" 66 | db { 67 | url = "jdbc:postgresql://127.0.0.1/slickdemo" 68 | driver = "org.postgresql.Driver" 69 | connectionPool = HikariCP 70 | user = "slick" 71 | password = "123" 72 | numThreads = 10 73 | maxConnections = 12 74 | minConnections = 4 75 | } 76 | } 77 | 78 | postgressdb = { 79 | dataSourceClass = "org.postgresql.ds.PGSimpleDataSource" 80 | properties = { 81 | databaseName = "slickdemo" 82 | user = "slick" 83 | password = "123" 84 | } 85 | connectionPool = HikariCP 86 | numThreads = 10 87 | maxConnections = 12 88 | minConnections = 4 89 | } 90 | 91 | mssql { 92 | driver = "com.typesafe.slick.driver.ms.SQLServerDriver$" 93 | db { 94 | url = "jdbc:sqlserver://host:port" 95 | driver = com.microsoft.sqlserver.jdbc.SQLServerDriver 96 | connectionTimeout = 30 second 97 | connectionPool = HikariCP 98 | user = "slick" 99 | password = "123" 100 | numThreads = 10 101 | maxConnections = 12 102 | minConnections = 4 103 | keepAliveConnection = true 104 | } 105 | } 106 | 107 | tsql { 108 | driver = "slick.driver.H2Driver$" 109 | db = ${h2mem} 110 | } -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdapipes/FDANodes.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdapipes 2 | import com.bayakala.funda._ 3 | import FDAJoints._ 4 | 5 | /** methods to run an user defined function*/ 6 | object FDANodes { //作业节点工作方法 7 | /** 8 | * returns state of next worknode. using fs2 Handle of Pull object, 9 | * take the next element and apply function task and determine new state of stream 10 | * @param task user defined function: ROW => Option[List[ROW]] 11 | * returns an Option[List[ROW]]] value signifying movement downstream 12 | * as follows: 13 | * Some(Nil) : skip sending the current row 14 | * Some(List(r1,r2...)): send r1,r2... downstream 15 | * None : halt stream, end of process 16 | * @tparam ROW row type: FDAROW or FDAActionROW 17 | * @return new state of stream 18 | */ 19 | def fda_execUserTask[ROW](task: FDAUserTask[ROW]): FDAWorkNode[ROW] = { 20 | def go: FDAValve[ROW] => FDAPipeJoint[ROW] = h => { 21 | h.receive1Option { 22 | case Some((r, h)) => task(r) match { 23 | case Some(lx) => lx match { 24 | case Nil => go(h) 25 | case _ => fda_pushRows(lx) >> go(h) 26 | } 27 | case None => {task(FDANullRow.asInstanceOf[ROW]); fda_halt} 28 | } 29 | case None => {task(FDANullRow.asInstanceOf[ROW]); fda_halt} 30 | } 31 | } 32 | in => in.pull(go) 33 | } 34 | /** 35 | * returns state of next worknode and some aggregation defined inside user function. 36 | * execute user defined function with internal aggregation mechanism by means of 37 | * functional state transition style of passing in state and return new state. 38 | * take in current aggregation and next row, apply user function on both 39 | * and determine new state of stream 40 | * @param aggr user selected type of aggregation such as Int, (Int,Int) ... 41 | * @param task user defined function: (AGGR,ROW) => (AGGR,Option[List[ROW]]) 42 | * take in current aggregation and row, 43 | * and return new aggregation and Option[List[ROW]] with meaning of: 44 | * Some(Nil) : skip sending the current row 45 | * Some(List(r1,r2...)): send r1,r2... downstream 46 | * None : halt stream, end of process 47 | * @tparam AGGR type of aggr 48 | * @tparam ROW type of row 49 | * @return new state of stream 50 | */ 51 | def fda_aggregate[AGGR,ROW](aggr: AGGR, task: FDAAggrTask[AGGR,ROW]): FDAWorkNode[ROW] = { 52 | def go(acc: AGGR): FDAValve[ROW] => FDAPipeJoint[ROW] = h => { 53 | h.receive1Option { 54 | case Some((r, h)) => task(acc,r) match { 55 | case (a,Some(lx)) => lx match { 56 | case Nil => go(a)(h) 57 | case _ => fda_pushRows(lx) >> go(a)(h) 58 | } 59 | case (a,None) => {task(a,FDANullRow.asInstanceOf[ROW]); fda_halt} 60 | } 61 | case None => {task(acc,FDANullRow.asInstanceOf[ROW]); fda_halt} 62 | } 63 | } 64 | in => in.pull(go(aggr)) 65 | } 66 | 67 | } 68 | -------------------------------------------------------------------------------- /src/main/scala/examples/ExceptionsAndFinalizers.scala: -------------------------------------------------------------------------------- 1 | package examples 2 | import slick.jdbc.H2Profile.api._ 3 | import com.bayakala.funda.samples.SlickModels._ 4 | import com.bayakala.funda._ 5 | import api._ 6 | import scala.language.implicitConversions 7 | 8 | object ExceptionsAndFinalizers extends App { 9 | 10 | val db = Database.forConfig("h2db") 11 | implicit def toState(row: StateTable#TableElementType) = 12 | StateModel(row.id,row.name) 13 | val viewLoader = FDAViewLoader(slick.jdbc.H2Profile)(toState _) 14 | val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toState _) 15 | 16 | val stateSeq = viewLoader.fda_typedRows(StateQuery.result)(db).toSeq 17 | val viewState = fda_staticSource(stateSeq)(println("***Finally*** the end of viewState!!!")) 18 | val streamState = streamLoader.fda_typedStream(StateQuery.result)(db)(64,64)(println("***Finally*** the end of streamState!!!"))() 19 | 20 | /* 21 | viewState.startRun 22 | viewState.take(2).startRun 23 | streamState.startRun 24 | streamState.take(3).startRun 25 | // ***Finally*** the end of viewState!!! 26 | // ***Finally*** the end of viewState!!! 27 | // ***Finally*** the end of streamState!!! 28 | // ***Finally*** the end of streamState!!! 29 | */ 30 | 31 | 32 | 33 | def trackRows: FDAUserTask[FDAROW] = row => { 34 | row match { 35 | case m@StateModel(id,name) => 36 | println(s"State: $id $name") 37 | println( "----------------") 38 | fda_next(m) 39 | case DivideZeroError(msg, e) => //error row 40 | println(s"***Error:$msg***") 41 | fda_skip 42 | case m@_ => fda_next(m) 43 | } 44 | } 45 | 46 | def errorRow: FDAUserTask[FDAROW] = row => { 47 | row match { 48 | case StateModel(id,name) => 49 | val idx = id / (id - 3) 50 | fda_next(StateModel(idx,name)) 51 | case m@_ => fda_next(m) 52 | } 53 | } 54 | 55 | case class DivideZeroError(msg: String, e: Exception) extends FDAROW 56 | def catchError: FDAUserTask[FDAROW] = row => { 57 | row match { 58 | case StateModel(id,name) => 59 | try { 60 | val idx = id / (id - 3) 61 | fda_next(StateModel(idx, name)) 62 | } catch { 63 | case e: Exception => //pass an error row 64 | fda_next(DivideZeroError(s"Divide by zero excption at ${id}",e)) 65 | } 66 | case m@_ => fda_next(m) 67 | } 68 | } 69 | 70 | 71 | 72 | /* 73 | streamState.appendTask(errorRow).appendTask(trackRows).startRun 74 | // State: 0 Alabama 75 | // ---------------- 76 | // State: -2 Alaska 77 | // ---------------- 78 | // Exception in thread "main" java.lang.ArithmeticException: / by zero 79 | // at examples.ExceptionsAndFinalizers$$anonfun$errorRow$1.apply(ExceptionsAndFinalizers.scala:46) 80 | // ... 81 | // at java.lang.Thread.run(Thread.java:745) 82 | // ***Finally*** the end of streamState!!! 83 | */ 84 | /* 85 | val v = viewState.appendTask(errorRow).appendTask(trackRows) 86 | val v1 = v.onError {case e: Exception => println(s"Caught Error in viewState!!![${e.getMessage}]"); fda_appendRow(FDANullRow)} 87 | v1.startRun 88 | 89 | val s = streamState.appendTask(errorRow).appendTask(trackRows) 90 | val s1 = s.onError {case e: Exception => println(s"Caught Error in streamState!!![${e.getMessage}]"); fda_appendRow(FDANullRow)} 91 | s1.startRun 92 | */ 93 | 94 | val s = streamState.take(5).appendTask(catchError).appendTask(trackRows) 95 | val s1 = s.onError {case e: Exception => println(s"Caught Error in streamState!!![${e.getMessage}]"); fda_appendRow(FDANullRow)} 96 | s1.startRun 97 | 98 | 99 | 100 | } 101 | -------------------------------------------------------------------------------- /src/main/scala/samples/SlickModels.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.samples 2 | 3 | import slick.jdbc.H2Profile.api._ 4 | import com.bayakala.funda._ 5 | 6 | object SlickModels { 7 | 8 | //表字段对应模版 9 | case class AQMRawModel(mid: String 10 | , mtype: String 11 | , state: String 12 | , fips: String 13 | , county: String 14 | , year: String 15 | , value: String) extends FDAROW 16 | 17 | //表结构: 定义字段类型, * 代表结果集字段 18 | class AQMRawTable(tag: Tag) extends Table[AQMRawModel](tag, "AIRQM") { 19 | def mid = column[String]("MEASUREID") 20 | def mtype = column[String]("MEASURETYPE") 21 | def state = column[String]("STATENAME") 22 | def fips = column[String]("COUNTYFIPS") 23 | def county = column[String]("COUNTYNAME") 24 | def year = column[String]("REPORTYEAR") 25 | def value = column[String]("VALUE") 26 | 27 | def * = (mid,mtype,state,fips,county,year,value) <> (AQMRawModel.tupled, AQMRawModel.unapply) 28 | } 29 | 30 | //库表实例 31 | val AQMRawQuery = TableQuery[AQMRawTable] 32 | 33 | case class AQMRPTModel(rid: Long 34 | , mid: Int 35 | , state: String 36 | , county: String 37 | , year: Int 38 | , value: Int 39 | , total: Int 40 | , valid: Boolean) extends FDAROW 41 | 42 | class AQMRPTTable(tag: Tag) extends Table[AQMRPTModel](tag, "AQMRPT") { 43 | def rid = column[Long]("ROWID",O.AutoInc,O.PrimaryKey) 44 | def mid = column[Int]("MEASUREID") 45 | def state = column[String]("STATENAME",O.Length(32)) 46 | def county = column[String]("COUNTYNAME",O.Length(32)) 47 | def year = column[Int]("REPORTYEAR") 48 | def value = column[Int]("VALUE") 49 | def total = column[Int]("TOTAL") 50 | def valid = column[Boolean]("VALID") 51 | 52 | def * = (rid,mid,state,county,year,value,total,valid) <> (AQMRPTModel.tupled, AQMRPTModel.unapply) 53 | } 54 | 55 | 56 | val AQMRPTQuery = TableQuery[AQMRPTTable] 57 | 58 | case class StateModel(id: Int, name: String) extends FDAROW 59 | class StateTable(tag: Tag) extends Table[StateModel](tag,"STATE") { 60 | def id = column[Int]("ID",O.AutoInc,O.PrimaryKey) 61 | def name = column[String]("NAME",O.Length(32)) 62 | def * = (id,name)<>(StateModel.tupled,StateModel.unapply) 63 | } 64 | val StateQuery = TableQuery[StateTable] 65 | 66 | case class CountyModel(id: Int, name: String) extends FDAROW 67 | case class CountyTable(tag: Tag) extends Table[CountyModel](tag,"COUNTY") { 68 | def id = column[Int]("ID",O.AutoInc,O.PrimaryKey) 69 | def name = column[String]("NAME",O.Length(64)) 70 | def * = (id,name)<>(CountyModel.tupled,CountyModel.unapply) 71 | } 72 | val CountyQuery = TableQuery[CountyTable] 73 | 74 | case class NORMAQMModel(rid: Long 75 | , mid: Int 76 | , state: Int 77 | , county: Int 78 | , year: Int 79 | , value: Int 80 | , average: Int 81 | ) extends FDAROW 82 | 83 | class NORMAQMTable(tag: Tag) extends Table[NORMAQMModel](tag, "NORMAQM") { 84 | def rid = column[Long]("ROWID",O.AutoInc,O.PrimaryKey) 85 | def mid = column[Int]("MEASUREID") 86 | def state = column[Int]("STATID") 87 | def county = column[Int]("COUNTYID") 88 | def year = column[Int]("REPORTYEAR") 89 | def value = column[Int]("VALUE") 90 | def average = column[Int]("AVG") 91 | 92 | def * = (rid,mid,state,county,year,value,average) <> (NORMAQMModel.tupled, NORMAQMModel.unapply) 93 | } 94 | 95 | 96 | val NORMAQMQuery = TableQuery[NORMAQMTable] 97 | 98 | 99 | } -------------------------------------------------------------------------------- /src/main/scala/examples/StrongTypedSource.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.examples 2 | import slick.jdbc.H2Profile.api._ 3 | import scala.language.implicitConversions 4 | import com.bayakala.funda._ 5 | import api._ 6 | import com.bayakala.funda.samples._ 7 | 8 | object StrongTypedSource extends App { 9 | 10 | val aqmraw = SlickModels.AQMRawQuery 11 | 12 | val db = Database.forConfig("h2db") 13 | // aqmQuery.result returns Seq[(String,String,String,String)] 14 | val aqmQuery = aqmraw.map {r => (r.year,r.state,r.county,r.value)} 15 | // user designed strong typed resultset type. must extend FDAROW 16 | case class TypedRow(year: String, state: String, county: String, value: String) extends FDAROW 17 | // strong typed resultset conversion function. declared implicit to remind during compilation 18 | implicit def toTypedRow(row: (String,String,String,String)): TypedRow = 19 | TypedRow(row._1,row._2,row._3,row._4) 20 | // loader to read from database and convert result collection to strong typed collection 21 | val viewLoader = FDAViewLoader(slick.jdbc.H2Profile)(toTypedRow _) 22 | val dataSeq = viewLoader.fda_typedRows(aqmQuery.result)(db).toSeq 23 | // turn Seq collection into fs2 stream 24 | val aqmStream = fda_staticSource(dataSeq)() 25 | // now access fields in the strong typed resultset 26 | def showRecord: FDAUserTask[FDAROW] = row => { 27 | row match { 28 | case qmr: TypedRow => 29 | println(s"州名: ${qmr.state}") 30 | println(s"县名：${qmr.county}") 31 | println(s"年份：${qmr.year}") 32 | println(s"取值：${qmr.value}") 33 | println("-------------") 34 | fda_skip 35 | case _ => fda_skip 36 | } 37 | } 38 | // use stream combinators with field names 39 | aqmStream.filter{r => r.year > "1999"}.take(3).appendTask(showRecord).startRun 40 | 41 | val allState = aqmraw.map(_.state) 42 | //no converter to help type inference. must provide type parameters explicitly 43 | val stateLoader = FDAViewLoader[String,String](slick.jdbc.H2Profile)() 44 | val stateSeq = stateLoader.fda_plainRows(allState.distinct.result)(db).toSeq 45 | //constructed a Stream[Task,String] 46 | val stateStream = fda_staticSource(stateSeq)() 47 | //strong typed row type. must extend FDAROW 48 | case class StateRow(state: String) extends FDAROW 49 | def showState: FDAUserTask[FDAROW] = row => { 50 | row match { 51 | case StateRow(sname) => 52 | println(s"州名称：$sname") 53 | fda_skip 54 | case _ => fda_skip 55 | } 56 | } 57 | 58 | //first convert to StateRows to turn Stream[Task,FDAROW] typed stream 59 | stateStream.map{s => StateRow(s)} 60 | .filter{r => r.state > "Alabama"}.take(3) 61 | .appendTask(showState).startRun 62 | 63 | 64 | 65 | //KillSwitch.killNow 66 | object killer extends Fs2Terminator 67 | 68 | val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toTypedRow _) 69 | val streamSource = streamLoader.fda_typedStream(aqmQuery.result)(db)(512,16)()(killer) 70 | // streamSource.filter{r => r.year > "1999"}.appendTask(showRecord).startRun 71 | streamSource 72 | .map {row => row match { 73 | case qmr: TypedRow if (qmr.value.toString == "5") => 74 | killer.stopASAP 75 | qmr 76 | case _ => row }} 77 | .appendTask(showRecord) 78 | .startRun 79 | 80 | 81 | val stateStreamLoader = FDAStreamLoader[String,String](slick.jdbc.H2Profile)() 82 | val stateStreamSource = stateStreamLoader.fda_plainStream(allState.distinct.result)(db)(512,512,30)() 83 | 84 | //first convert to StateRows to turn Stream[Task,FDAROW] typed stream 85 | stateStreamSource.map{s => StateRow(s)} 86 | .filter{r => r.state > "Alabama"}.take(3) 87 | .appendTask(showState).startRun 88 | 89 | val akkaStreamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toTypedRow _) 90 | val akkaStreamSource = akkaStreamLoader.fda_akkaTypedStream(aqmQuery.result)(db)(512,512,30)()() 91 | akkaStreamSource //.filter{r => r.year > "1999"} 92 | .appendTask(showRecord).startRun 93 | 94 | val akkaStreamSource2 = akkaStreamLoader.fda_akkaTypedStream(aqmQuery.result)(db)(512,1)()() 95 | akkaStreamSource2 96 | .map {row => row match { 97 | case qmr: TypedRow if (qmr.value.toString == "5") => 98 | AkkaKillSwitch.stopASAP 99 | qmr 100 | case _ => row }} 101 | .appendTask(showRecord) 102 | .startRun 103 | 104 | 105 | } -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdapars/FDAPars.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdapars 2 | 3 | import fs2._ 4 | import com.bayakala.funda._ 5 | import com.bayakala.funda.fdapipes.FDAJoints._ 6 | /** operation methods for parallel computation */ 7 | trait FDAPars { 8 | /** 9 | * return a single stream of combined data rows. 10 | * reading from a list of many streams defined in sources concurrently. 11 | * @example {{{ 12 | * //3 separate streams to extract county names from the same database table AQMRPT 13 | * val countiesA_KStream = countyLoader.fda_typedStream(qryCountiesA_K.result)(db_b)(64,64)() 14 | * val countiesK_PStream = countyLoader.fda_typedStream(qryCountiesK_P.result)(db_b)(64,64)() 15 | * val countiesP_ZStream = countyLoader.fda_typedStream(qryCountiesP_Z.result)(db_b)(64,64)() 16 | * 17 | * //obtain a combined stream with parallel loading with max of 4 open computation 18 | * val combinedStream = fda_par_load(statesStream,countiesA_KStream,countiesK_PStream,countiesP_ZStream)(4) 19 | * }}} 20 | * @param sources list of many individual streams 21 | * @param maxOpen maximum number of concurrently open computation. 22 | * could be a number near and less than the cpu cores 23 | * @return a combined data stream with nondeterministic element order 24 | */ 25 | def fda_par_load(sources: FDAPipeLine[FDAROW]*)(maxOpen: Int): FDAPipeLine[FDAROW] = { 26 | concurrent.join(maxOpen)(Stream(sources: _*)) 27 | } 28 | /** 29 | * run user defined function on a single stream many ways in parallel. 30 | * return a single stream as result of user function application. 31 | * @example {{{ 32 | * //turn getIdsThenInsertAction into parallel task 33 | * val parTasks = source.toPar(getIdsThenInsertAction) 34 | * //runPar to produce a new stream 35 | * val actionStream =fda_runPar(parTasks)(4) 36 | * }}} 37 | * @param parTask user defined function turned into par style by toPar 38 | * @param maxOpen maximum number of concurrently open computation. 39 | * could be a number near and less than the cpu cores 40 | * @return a single stream with nondeterministic element order 41 | */ 42 | def fda_runPar(parTask: FDAParTask)(maxOpen: Int) = 43 | concurrent.join(maxOpen)(parTask).through(fda_afterPar) 44 | 45 | /** 46 | * load many sources in parallel. 47 | * return a single combined stream with non-deterministic row order. 48 | * @example {{{ 49 | * //loading rows with year yr 50 | * def loadRowsInYear(yr: Int) = { 51 | * //a new query 52 | * val query = AQMRPTQuery.filter(row => row.year === yr) 53 | * //reuse same loader 54 | * AQMRPTLoader.fda_typedStream(query.result)(db)(256, 256)(println(s"End of stream ${yr}!!!!!!")) 55 | * } 56 | * 57 | * //loading rows by year 58 | * def loadRowsByYear: FDASourceLoader = row => { 59 | * row match { 60 | * case Years(y) => loadRowsInYear(y) //produce stream of the year 61 | * case _ => fda_appendRow(FDANullRow) 62 | * } 63 | * } 64 | * 65 | * //get parallel source constructor 66 | * val parSource = yearStream.toParSource(loadRowsByYear) 67 | * //produce a stream from parallel sources 68 | * val source = fda_par_source(parSource)(4) 69 | * }}} 70 | * @param parSource user defined source constructor turned into 71 | * par style by toParSource 72 | * @param maxOpen maximum number of concurrently open computation. 73 | * could be a number near and less than the cpu cores 74 | * @return a single stream with nondeterministic element order 75 | */ 76 | def fda_par_source(parSource: FDAParSource)(maxOpen: Int) = 77 | concurrent.join(maxOpen)(parSource) 78 | 79 | 80 | /** flatten the Option[List[FDAROW]]] type returned by fda_runPar to plain FDAROW */ 81 | private def fda_afterPar: Pipe[Task,Option[List[FDAROW]],FDAROW] = { 82 | def go: Handle[Task,Option[List[FDAROW]]] => FDAPipeJoint[FDAROW] = h => { 83 | h.receive1Option { 84 | case Some((r, h)) => r match { 85 | case Some(xr) => 86 | fda_pushRows(xr) >> go(h) 87 | case None => go(h) 88 | } 89 | case None => fda_halt 90 | } 91 | } 92 | in => in.pull(go) 93 | } 94 | 95 | } 96 | 97 | /** 98 | * for global imports 99 | */ 100 | object FDAPars extends FDAPars -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdarows/FDARowOperator.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdarows 2 | import com.bayakala.funda._ 3 | import scala.concurrent.duration._ 4 | import scala.concurrent.Await 5 | import slick.jdbc.JdbcProfile 6 | 7 | /** 8 | * loading of data rows and running of action rows 9 | */ 10 | trait FDARowOperator { 11 | 12 | /** 13 | * read view, an entire collection of data as result of Slick DBIO action 14 | * convert collection data type if required 15 | * run FDAAction row as Slick run(DBIOAction) 16 | * @param slickProfile Slick jdbcprofile such as 'slick.jdbc.H2Profile' 17 | * @param convert a defined implicit type conversion function. 18 | * from SOURCE type to TARGET type, set to null if not required 19 | * @tparam SOURCE source type, result type of DBIOAction, most likely a tuple type 20 | * @tparam TARGET final converted type, most likely a case class type 21 | */ 22 | class FDAViewLoader[SOURCE, TARGET](slickProfile: JdbcProfile, convert: SOURCE => TARGET) { 23 | 24 | import slickProfile.api._ 25 | 26 | /** 27 | * return collection of TARGET type data. 28 | * run slickAction and apply convert result from collection of SOURCE type 29 | * to collection of TARGET type 30 | * @example {{{ 31 | * val viewLoader = FDAViewLoader(slick.jdbc.H2Profile)(toTypedRow _) 32 | * val dataSeq = viewLoader.fda_typedRows(aqmQuery.result)(db).toSeq 33 | * }}} 34 | * @param slickAction Slick DBIOAction for query data results 35 | * @param slickDB Slick database object 36 | * @param converter just a measure to guarantee conversion function is defined 37 | * when this function is used there has to be a converter defined 38 | * implicitly in compile time 39 | * @return a collection of TARGET type data 40 | */ 41 | def fda_typedRows(slickAction: DBIO[Iterable[SOURCE]])(slickDB: Database)( 42 | implicit converter: SOURCE => TARGET): Iterable[TARGET] = { 43 | Await.result(slickDB.run(slickAction), Duration.Inf).map(raw => convert(raw)) 44 | } 45 | 46 | /** 47 | * load view without type conversion of result collection data type 48 | * @example {{{ 49 | * val viewLoader = FDAViewLoader(slick.jdbc.H2Profile)() 50 | * val dataSeq = viewLoader.fda_plainRows(aqmQuery.result)(db).toSeq 51 | * }}} 52 | * @param slickAction Slick DBIOAction for query data results 53 | * @param slickDB Slick database object 54 | * @return a collection of SOURCE type data 55 | */ 56 | def fda_plainRows(slickAction: DBIO[Iterable[SOURCE]])(slickDB: Database): Iterable[SOURCE] = { 57 | Await.result(slickDB.run(slickAction), Duration.Inf) 58 | } 59 | } 60 | 61 | /** 62 | * constructor given slickprofile and converter function 63 | */ 64 | object FDAViewLoader{ 65 | /** 66 | * FDAViewLoader constructor 67 | * @example {{{ 68 | * val viewLoader = FDAViewLoader(slick.jdbc.H2Profile)(toTypedRow _) 69 | * val untypedLoader = FDAViewLoader(slick.jdbc.H2Profile)() 70 | * }}} 71 | * @param slickProfile Slick jdbcprofile such as 'slick.jdbc.H2Profile' 72 | * @param converter type conversion function. from SOURCE type to TARGET type 73 | * @tparam SOURCE source type, result type of DBIOAction, most likely a tuple type 74 | * @tparam TARGET final converted type, most likely a case class type 75 | * @return a new FDAViewLoader object 76 | */ 77 | def apply[SOURCE, TARGET](slickProfile: JdbcProfile)( 78 | converter: SOURCE => TARGET = null): FDAViewLoader[SOURCE, TARGET] = 79 | new FDAViewLoader[SOURCE, TARGET](slickProfile, converter) 80 | } 81 | 82 | /** 83 | * method to run action rows of type FDAAction 84 | * @param slickProfile Slick jdbcprofile such as 'slick.jdbc.H2Profile' 85 | */ 86 | class FDAActionRunner(slickProfile: JdbcProfile) { 87 | 88 | import slickProfile.api._ 89 | 90 | /** 91 | * run a Slick DBIOAction action. 92 | * @param action a Slick DBIOAction 93 | * @param slickDB Slick database object 94 | * @return rows affected 95 | */ 96 | def fda_execAction(action: FDAAction)(slickDB: Database): Int = 97 | Await.result(slickDB.run(action), Duration.Inf) 98 | } 99 | 100 | /** 101 | * constructor given jdbcprofile 102 | */ 103 | object FDAActionRunner { 104 | /** 105 | * construct a FDAActionRunner object given slickProfile 106 | * @param slickProfile Slick jdbcprofile such as 'slick.jdbc.H2Profile' 107 | * @return a new FDAActionRunner object 108 | */ 109 | def apply(slickProfile: JdbcProfile): FDAActionRunner = new FDAActionRunner(slickProfile) 110 | } 111 | } 112 | 113 | /** 114 | * for global imports 115 | */ 116 | object FDARowOperator extends FDARowOperator 117 | -------------------------------------------------------------------------------- /src/main/resources/logback.xml: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | *** \(%logger{30}\)%green(%X{debugId}) %msg%n 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 | 46 | 47 | 48 | 49 | 50 | 51 | 52 | 53 | 54 | 55 | 56 | 57 | -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdapipes/FDAPipes.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdapipes 2 | import com.bayakala.funda._ 3 | 4 | trait FDAPipes { 5 | /** 6 | * returns state of next worknode. using fs2 Handle of Pull object, 7 | * take the next element and apply function task and determine new state of stream 8 | * @param task user defined function: ROW => Option[List[ROW]] 9 | * returns an Option[List[ROW]]] value signifying movement downstream 10 | * as follows: 11 | * Some(Nil) : skip sending the current row 12 | * Some(List(r1,r2...)): send r1,r2... downstream 13 | * None : halt stream, end of process 14 | * @tparam ROW row type: FDAROW or FDAActionROW 15 | * @return new state of stream 16 | */ 17 | private def fda_execUserTask[ROW](task: FDAUserTask[ROW]): FDAWorkNode[ROW] = FDANodes.fda_execUserTask(task) 18 | /** 19 | * returns state of next worknode and some aggregation defined inside user function. 20 | * execute user defined function with internal aggregation mechanism by means of 21 | * functional state transition style of passing in state and return new state. 22 | * take in current aggregation and next row, apply user function on both 23 | * and determine new state of stream 24 | * @param aggr user selected type of aggregation such as Int, (Int,Int) ... 25 | * @param task user defined function: (AGGR,ROW) => (AGGR,Option[List[ROW]]) 26 | * take in current aggregation and row, 27 | * and return new aggregation and Option[List[ROW]] with meaning of: 28 | * Some(Nil) : skip sending the current row 29 | * Some(List(r1,r2...)): send r1,r2... downstream 30 | * None : halt stream, end of process 31 | * @tparam AGGR type of aggr 32 | * @tparam ROW type of row 33 | * @return new state of stream 34 | */ 35 | private def fda_aggregateTask[AGGR,ROW](aggr: AGGR, task: FDAAggrTask[AGGR,ROW]): FDAWorkNode[ROW] = FDANodes.fda_aggregate(aggr,task) 36 | /** 37 | * send row to Pull 38 | * @param row row to be sent 39 | * @tparam ROW type of row 40 | * @return new state of Pull 41 | */ 42 | private def fda_pushRow[ROW](row: ROW): FDAPipeJoint[ROW] = FDAJoints.fda_pushRow(row) 43 | /** 44 | * send list of rows to Pull 45 | * @param rows list of rows to be sent 46 | * @tparam ROW type of target row 47 | * @return new state of Pull 48 | */ 49 | private def fda_pushRows[ROW](rows: List[ROW]): FDAPipeJoint[ROW] = FDAJoints.fda_pushRows(rows) 50 | /** 51 | * end output to Pull 52 | * @return end of Pull 53 | */ 54 | private def fda_halt = FDAJoints.fda_halt 55 | /** 56 | * skip current row by sending a Some(Nil) 57 | * @example {{{ 58 | * def trackRows: FDAUserTask[FDAROW] = row => { 59 | * row match { 60 | * case m@StateModel(id,name) => 61 | * println(s"State: $id $name") 62 | * println( "----------------") 63 | * fda_next(m) 64 | * case DivideZeroError(msg, e) => //error row 65 | * println(s"***Error:$msg***") 66 | * fda_skip 67 | * case m@_ => fda_next(m) 68 | * } 69 | * } 70 | * }}} 71 | * @return Some(Nil) to signify skipping current row 72 | */ 73 | def fda_skip[ROW]: Option[List[ROW]] = FDAValves.fda_skip[ROW] 74 | /** 75 | * send row downstream 76 | * @example {{{ 77 | * def trackRows: FDAUserTask[FDAROW] = row => { 78 | * row match { 79 | * case m@StateModel(id,name) => 80 | * println(s"State: $id $name") 81 | * println( "----------------") 82 | * fda_next(m) 83 | * case DivideZeroError(msg, e) => //error row 84 | * println(s"***Error:$msg***") 85 | * fda_skip 86 | * case m@_ => fda_next(m) 87 | * } 88 | * } 89 | * }}} 90 | * @param row target row 91 | * @tparam ROW type of target row 92 | * @return a single row to be sent downstream 93 | */ 94 | def fda_next[ROW](row: ROW): Option[List[ROW]] = FDAValves.fda_next(row) 95 | /** 96 | * send a list of rows downstream in a chunk 97 | * @example {{{ 98 | * def trackRows: FDAUserTask[FDAROW] = row => { 99 | * row match { 100 | * case m@StateModel(id,name) => 101 | * println(s"State: $id $name") 102 | * println( "----------------") 103 | * fda_next(m) 104 | * case DivideZeroError(msg, e) => //error row 105 | * println(s"***Error:$msg***") 106 | * fda_skip 107 | * case m@_ => fda_next(List(m,m)) //send double row 108 | * } 109 | * } 110 | * }}} 111 | * @param lr list of rows to be sent 112 | * @tparam ROW type of target row 113 | * @return a list of many rows 114 | */ 115 | def fda_next[ROW](lr: List[ROW]): Option[List[ROW]] = FDAValves.fda_next(lr) 116 | /** 117 | * halt the current stream 118 | * @example {{{ 119 | * //loading rows by year 120 | * def loadRowsByYear: FDASourceLoader = row => { 121 | * row match { 122 | * case Years(y) => loadRowsInYear(y) //produce stream of the year 123 | * case _ => //something wrong break and terminate this process 124 | * fda_break 125 | * } 126 | * } 127 | * }}} 128 | * @return a None indicating end of current stream process 129 | */ 130 | def fda_break = FDAValves.fda_break 131 | } 132 | 133 | /** 134 | * for global imports 135 | */ 136 | object FDAPipes extends FDAPipes -------------------------------------------------------------------------------- /src/main/scala/examples/ParallelLoading.scala: -------------------------------------------------------------------------------- 1 | package examples 2 | import com.bayakala.funda._ 3 | import api._ 4 | import scala.language.implicitConversions 5 | import slick.jdbc.H2Profile.api._ 6 | import scala.concurrent.duration._ 7 | import scala.concurrent.{Await, Future} 8 | import scala.util.{Failure, Success} 9 | import com.bayakala.funda.samples.SlickModels._ 10 | import scala.concurrent.ExecutionContext.Implicits.global 11 | 12 | object ParallelLoading extends App { 13 | 14 | //assume two distinct db objects 15 | val db_a = Database.forConfig("h2db") 16 | //another db object 17 | val db_b = Database.forConfig("h2db") 18 | 19 | 20 | //create STATE table 21 | val actionCreateState = StateQuery.schema.create 22 | val futCreateState = db_a.run(actionCreateState).andThen { 23 | case Success(_) => println("State Table created successfully!") 24 | case Failure(e) => println(s"State Table may exist already! Error: ${e.getMessage}") 25 | } 26 | //would carry on even fail to create table 27 | Await.ready(futCreateState,Duration.Inf) 28 | 29 | //create COUNTY table 30 | val actionCreateCounty = CountyQuery.schema.create 31 | val futCreateCounty = db_a.run(actionCreateCounty).andThen { 32 | case Success(_) => println("County Table created successfully!") 33 | case Failure(e) => println(s"County Table may exist already! Error: ${e.getMessage}") 34 | } 35 | //would carry on even fail to create table 36 | Await.ready(futCreateCounty,Duration.Inf) 37 | 38 | //define query for extracting State names from AQMRPT 39 | val qryStates = AQMRPTQuery.map(_.state).distinct.sorted // .distinctOn(r => r) 40 | case class States(name: String) extends FDAROW 41 | implicit def toStates(row: String) = States(row) 42 | val stateLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toStates _) 43 | val statesStream = stateLoader.fda_typedStream(qryStates.result)(db_a)(64,64)()() 44 | 45 | 46 | //define query for extracting County names from AQMRPT in separate chunks 47 | //query with state name >A and (r.state.toUpperCase > "A" && 49 | r.state.toUpperCase < "K")).map(r => (r.state,r.county)) 50 | .distinctOn(r => (r._1,r._2)) 51 | .sortBy(r => (r._1,r._2)) 52 | 53 | //query with state name >K and

(r.state.toUpperCase > "K" && 55 | r.state.toUpperCase < "P")).map(r => (r.state,r.county)) 56 | .distinctOn(r => (r._1,r._2)) 57 | .sortBy(r => (r._1,r._2)) 58 | 59 | //query with state name >P 60 | val qryCountiesP_Z = AQMRPTQuery.filter(r => r.state.toUpperCase > "P") 61 | .map(r => (r.state,r.county)) 62 | .distinctOn(r => (r._1,r._2)) 63 | .sortBy(r => (r._1,r._2)) 64 | 65 | case class Counties(state: String, name: String) extends FDAROW 66 | implicit def toCounties(row: (String,String)) = Counties(row._1,row._2) 67 | val countyLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toCounties _) 68 | //3 separate streams to extract county names from the same database table AQMRPT 69 | val countiesA_KStream = countyLoader.fda_typedStream(qryCountiesA_K.result)(db_b)(64,64)()() 70 | val countiesK_PStream = countyLoader.fda_typedStream(qryCountiesK_P.result)(db_b)(64,64)()() 71 | val countiesP_ZStream = countyLoader.fda_typedStream(qryCountiesP_Z.result)(db_b)(64,64)()() 72 | 73 | //obtain a combined stream with parallel loading with max of 4 open computation 74 | val combinedStream = fda_par_load(statesStream,countiesA_KStream,countiesK_PStream,countiesP_ZStream)(4) 75 | 76 | 77 | //define separate rows for different actions 78 | case class StateActionRow(action: FDAAction) extends FDAROW 79 | case class CountyActionRow(action: FDAAction) extends FDAROW 80 | val actionRunner = FDAActionRunner(slick.jdbc.H2Profile) 81 | 82 | //user-task to catch rows of States type and transform them into db insert actions 83 | def processStates: FDAUserTask[FDAROW] = row => { 84 | row match { 85 | //catch states row and transform it into insert action 86 | case States(stateName) => //target row type 87 | println(s"State name: ${stateName}") 88 | val action = StateQuery += StateModel(0,stateName) 89 | fda_next(StateActionRow(action)) 90 | case others@ _ => //pass other types to next user-defined-tasks 91 | fda_next(others) 92 | } 93 | } 94 | //user-task to catch rows of Counties type and transform them into db insert actions 95 | def processCounties: FDAUserTask[FDAROW] = row => { 96 | row match { 97 | //catch counties row and transform it into insert action 98 | case Counties(stateName,countyName) => //target row type 99 | println(s"County ${countyName} of ${stateName}") 100 | val action = CountyQuery += CountyModel(0,countyName+ " of "+stateName) 101 | fda_next(CountyActionRow(action)) 102 | case others@ _ => //pass other types to next user-defined-tasks 103 | fda_next(others) 104 | } 105 | } 106 | 107 | //user-task to catch States insert action rows and run them 108 | def runStateAction: FDAUserTask[FDAROW] = row => { 109 | row match { 110 | case StateActionRow(action) => //this is a state action row type 111 | println(s"runstate: ${action}") 112 | actionRunner.fda_execAction(action)(db_a) //run this query with db_a context 113 | fda_skip 114 | case others@ _ => //otherwise pass alone to next user-defined-tasks 115 | fda_next(others) 116 | } 117 | } 118 | 119 | //user-task to catch Counties insert action rows and run them 120 | def runCountyAction: FDAUserTask[FDAROW] = row => { 121 | row match { 122 | case CountyActionRow(action) => //this is a county action row type 123 | actionRunner.fda_execAction(action)(db_b) //run this query with db_b context 124 | fda_skip 125 | case others@ _ => //otherwise pass alone to next user-defined-tasks 126 | fda_next(others) 127 | } 128 | } 129 | 130 | 131 | 132 | def showRows: FDAUserTask[FDAROW] = row => { 133 | row match { 134 | case States(nm) => 135 | println("") 136 | println(s"State: $nm") 137 | println("************") 138 | fda_skip 139 | case Counties(s,c) => 140 | println("") 141 | println(s"County: $c") 142 | println(s"state of $s") 143 | println("------------") 144 | fda_skip 145 | case _ => fda_skip 146 | } 147 | } 148 | 149 | combinedStream.appendTask(processStates) 150 | .appendTask(processCounties) 151 | .appendTask(runStateAction) 152 | .appendTask(runCountyAction) 153 | .startRun 154 | 155 | } 156 | -------------------------------------------------------------------------------- /src/main/scala/examples/UserDefinedTask.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.examples 2 | import slick.jdbc.meta._ 3 | import scala.language.implicitConversions 4 | import scala.concurrent.ExecutionContext.Implicits.global 5 | import scala.concurrent.duration._ 6 | import scala.concurrent.{Await, Future} 7 | import scala.util.{Failure, Success} 8 | import slick.jdbc.H2Profile.api._ 9 | import com.bayakala.funda._ 10 | import api._ 11 | import com.bayakala.funda.samples.SlickModels._ 12 | 13 | object UserDefinedTasks extends App { 14 | 15 | 16 | val db = Database.forConfig("h2db") 17 | 18 | //drop original table schema 19 | val futVectorTables = db.run(MTable.getTables) 20 | 21 | val futDropTable = futVectorTables.flatMap{ tables => { 22 | val tableNames = tables.map(t => t.name.name) 23 | if (tableNames.contains(AQMRPTQuery.baseTableRow.tableName)) 24 | db.run(AQMRPTQuery.schema.drop) 25 | else Future(():Unit) 26 | } 27 | }.andThen { 28 | case Success(_) => println(s"Table ${AQMRPTQuery.baseTableRow.tableName} dropped successfully! ") 29 | case Failure(e) => println(s"Failed to drop Table ${AQMRPTQuery.baseTableRow.tableName}, it may not exist! Error: ${e.getMessage}") 30 | } 31 | Await.ready(futDropTable,Duration.Inf) 32 | 33 | //create new table to refine AQMRawTable 34 | val actionCreateTable = AQMRPTQuery.schema.create 35 | val futCreateTable = db.run(actionCreateTable).andThen { 36 | case Success(_) => println("Table created successfully!") 37 | case Failure(e) => println(s"Table may exist already! Error: ${e.getMessage}") 38 | } 39 | //would carry on even fail to create table 40 | Await.ready(futCreateTable,Duration.Inf) 41 | 42 | 43 | //truncate data, only available in slick 3.2.1 44 | val futTruncateTable = futVectorTables.flatMap{ tables => { 45 | val tableNames = tables.map(t => t.name.name) 46 | if (tableNames.contains(AQMRPTQuery.baseTableRow.tableName)) 47 | db.run(AQMRPTQuery.schema.truncate) 48 | else Future(():Unit) 49 | } 50 | }.andThen { 51 | case Success(_) => println(s"Table ${AQMRPTQuery.baseTableRow.tableName} truncated successfully!") 52 | case Failure(e) => println(s"Failed to truncate Table ${AQMRPTQuery.baseTableRow.tableName}! Error: ${e.getMessage}") 53 | } 54 | Await.ready(futDropTable,Duration.Inf) 55 | 56 | 57 | //load original table content 58 | //original table strong-typed-row 59 | case class AQMRaw(mid: String, state: String, 60 | county: String, year: String, value: String) extends FDAROW 61 | implicit def toAQMRaw(row: (String,String,String,String,String)) = 62 | AQMRaw(row._1,row._2,row._3,row._4,row._5) 63 | val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toAQMRaw _) 64 | // val queryAQMRaw = for { r <- AQMRawQuery } yield (r.mid,r.state,r.county,r.year,r.value) 65 | val queryAQMRaw = sql""" 66 | SELECT MEASUREID,STATENAME,COUNTYNAME,REPORTYEAR,VALUE FROM AIRQM 67 | """.as[(String,String,String,String,String)] 68 | 69 | val streamAQMRaw: FDAPipeLine[FDAROW] = streamLoader.fda_typedStream(queryAQMRaw)(db)(512,512)()() 70 | 71 | 72 | //filter out rows with inconvertible value strings and out of ranged value and year 73 | def filterRows: FDAUserTask[FDAROW] = row => { 74 | row match { 75 | case r: AQMRaw => { 76 | try { 77 | val yr = r.year.toInt 78 | val v = r.value.toInt 79 | val vlu = if ( v > 10 ) 10 else v 80 | val data = AQMRPTModel(0,r.mid.toInt,r.state,r.county,yr,vlu,0,true) 81 | if ((yr > 1960 && yr < 2018)) 82 | fda_next(data) //this row ok. pass downstream 83 | else 84 | fda_skip //filter out this row 85 | } catch { 86 | case e: Exception => 87 | fda_next(AQMRPTModel(0,r.mid.toInt,r.state,r.county,2000,0,0,false)) 88 | //pass a invalid row 89 | } 90 | } 91 | case _ => fda_skip //wrong type, skip 92 | } 93 | } 94 | 95 | //transform data to action for later execution 96 | def toAction: FDAUserTask[FDAROW] = row => { 97 | row match { 98 | case r: AQMRPTModel => 99 | val queryAction = AQMRPTQuery += r 100 | fda_next(FDAActionRow(queryAction)) 101 | case other @ _ => fda_next(other) 102 | } 103 | } 104 | 105 | //get a query runner and an action task 106 | val actionRunner = FDAActionRunner(slick.jdbc.H2Profile) 107 | def runActionRow: FDAUserTask[FDAROW] = action => { 108 | action match { 109 | case FDAActionRow(q) => actionRunner.fda_execAction(q)(db) 110 | fda_skip 111 | case _ => fda_skip 112 | } 113 | } 114 | 115 | 116 | //start the program 117 | val streamAllTasks = streamAQMRaw.appendTask(filterRows) 118 | .appendTask(toAction) 119 | .appendTask(runActionRow) 120 | 121 | val streamToRun = streamAllTasks.onError { case e: Exception => println("Error:"+e.getMessage); fda_appendRow(FDAErrorRow(new Exception(e))) } 122 | 123 | streamToRun.startRun 124 | 125 | //aggregate-task demo: get count and sum of value for each state and year 126 | val orderedAQMRPT = AQMRPTQuery.sortBy(r => (r.state,r.year)) 127 | //TableElementType conversion. must declare implicit 128 | implicit def toAQMRPT(row: AQMRPTTable#TableElementType) = 129 | AQMRPTModel(row.rid,row.mid,row.state,row.county,row.year,row.value,row.total,row.valid) 130 | val aqmrStreamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toAQMRPT _) 131 | val aqmrStream: FDAPipeLine[FDAROW] = aqmrStreamLoader.fda_typedStream(orderedAQMRPT.result)(db)(512,512)()() 132 | //user defined aggregator type. 133 | case class Accu(state: String, county: String, year: Int, count: Int, sumOfValue: Int) 134 | //user defined aggregation task 135 | def aggregateValue: FDAAggrTask[Accu,FDAROW] = (accu,row) => { 136 | row match { 137 | case aqmr: AQMRPTModel => 138 | if (accu.state == "" || (aqmr.state == accu.state && aqmr.year == accu.year)) 139 | //same condition: inc count and add sum, pass no row downstream 140 | (Accu(aqmr.state,aqmr.county,aqmr.year,accu.count+1, accu.sumOfValue+aqmr.value),fda_skip) 141 | else 142 | //reset accumulator, create a new aggregated row and pass downstream 143 | (Accu(aqmr.state,aqmr.county,aqmr.year,1, aqmr.value) 144 | ,fda_next(AQMRPTModel(0,9999,accu.state,accu.county,accu.year 145 | ,accu.count,accu.sumOfValue/accu.count,true))) 146 | case FDANullRow => 147 | //last row encountered. create and pass new aggregated row 148 | (Accu(accu.state,accu.county,accu.year,1, 0) 149 | ,fda_next(AQMRPTModel(0,9999,accu.state,accu.county,accu.year 150 | ,accu.count,accu.sumOfValue/accu.count,true))) 151 | //incorrect row type, do nothing 152 | case _ => (accu,fda_skip) 153 | } 154 | } 155 | 156 | 157 | aqmrStream.aggregateTask(Accu("","",0,0,0),aggregateValue) 158 | .appendTask(toAction) 159 | .appendTask(runActionRow) 160 | .startRun 161 | 162 | 163 | } 164 | -------------------------------------------------------------------------------- /src/main/scala/examples/ParallelTasks.scala: -------------------------------------------------------------------------------- 1 | package examples 2 | import slick.jdbc.meta._ 3 | import com.bayakala.funda._ 4 | import api._ 5 | import scala.language.implicitConversions 6 | import scala.concurrent.ExecutionContext.Implicits.global 7 | import scala.concurrent.duration._ 8 | import scala.concurrent.{Await, Future} 9 | import scala.util.{Failure, Success} 10 | import slick.jdbc.H2Profile.api._ 11 | import com.bayakala.funda.samples.SlickModels._ 12 | import fs2.Strategy 13 | 14 | object ParallelTasks extends App { 15 | 16 | val db = Database.forConfig("h2db") 17 | 18 | //drop original table schema 19 | val futVectorTables = db.run(MTable.getTables) 20 | 21 | val futDropTable = futVectorTables.flatMap{ tables => { 22 | val tableNames = tables.map(t => t.name.name) 23 | if (tableNames.contains(NORMAQMQuery.baseTableRow.tableName)) 24 | db.run(NORMAQMQuery.schema.drop) 25 | else Future(():Unit) 26 | } 27 | }.andThen { 28 | case Success(_) => println(s"Table ${NORMAQMQuery.baseTableRow.tableName} dropped successfully! ") 29 | case Failure(e) => println(s"Failed to drop Table ${NORMAQMQuery.baseTableRow.tableName}, it may not exist! Error: ${e.getMessage}") 30 | } 31 | Await.ready(futDropTable,Duration.Inf) 32 | 33 | //create new table to refine AQMRawTable 34 | val actionCreateTable = NORMAQMQuery.schema.create 35 | val futCreateTable = db.run(actionCreateTable).andThen { 36 | case Success(_) => println("Table created successfully!") 37 | case Failure(e) => println(s"Table may exist already! Error: ${e.getMessage}") 38 | } 39 | //would carry on even fail to create table 40 | Await.ready(futCreateTable,Duration.Inf) 41 | 42 | 43 | //truncate data, only available in slick 3.2.1 44 | val futTruncateTable = futVectorTables.flatMap{ tables => { 45 | val tableNames = tables.map(t => t.name.name) 46 | if (tableNames.contains(NORMAQMQuery.baseTableRow.tableName)) 47 | db.run(NORMAQMQuery.schema.truncate) 48 | else Future(():Unit) 49 | } 50 | }.andThen { 51 | case Success(_) => println(s"Table ${NORMAQMQuery.baseTableRow.tableName} truncated successfully!") 52 | case Failure(e) => println(s"Failed to truncate Table ${NORMAQMQuery.baseTableRow.tableName}! Error: ${e.getMessage}") 53 | } 54 | Await.ready(futDropTable,Duration.Inf) 55 | 56 | //a conceived task for the purpose of resource consumption 57 | //getting id with corresponding name from STATES table 58 | def getStateID(state: String): Int = { 59 | //create a stream for state id with state name 60 | implicit def toState(row: StateTable#TableElementType) = StateModel(row.id,row.name) 61 | val stateLoader = FDAViewLoader(slick.jdbc.H2Profile)(toState _) 62 | val stateSeq = stateLoader.fda_typedRows(StateQuery.result)(db).toSeq 63 | //constructed a Stream[Task,String] 64 | val stateStream = fda_staticSource(stateSeq)() 65 | var id = -1 66 | def getid: FDAUserTask[FDAROW] = row => { 67 | row match { 68 | case StateModel(stid,stname) => //target row type 69 | if (stname.contains(state)) { 70 | id = stid 71 | fda_break //exit 72 | } 73 | else fda_skip //take next row 74 | case _ => fda_skip 75 | } 76 | } 77 | stateStream.appendTask(getid).startRun 78 | id 79 | } 80 | //another conceived task for the purpose of resource consumption 81 | //getting id with corresponding names from COUNTIES table 82 | def getCountyID(state: String, county: String): Int = { 83 | //create a stream for county id with state name and county name 84 | implicit def toCounty(row: CountyTable#TableElementType) = CountyModel(row.id,row.name) 85 | val countyLoader = FDAViewLoader(slick.jdbc.H2Profile)(toCounty _) 86 | val countySeq = countyLoader.fda_typedRows(CountyQuery.result)(db).toSeq 87 | //constructed a Stream[Task,String] 88 | val countyStream = fda_staticSource(countySeq)() 89 | var id = -1 90 | def getid: FDAUserTask[FDAROW] = row => { 91 | row match { 92 | case CountyModel(cid,cname) => //target row type 93 | if (cname.contains(state) && cname.contains(county)) { 94 | id = cid 95 | fda_break //exit 96 | } 97 | else fda_skip //take next row 98 | case _ => fda_skip 99 | } 100 | } 101 | countyStream.appendTask(getid).startRun 102 | id 103 | } 104 | 105 | //original table listing 106 | implicit def toAQMRPT(row: AQMRPTTable#TableElementType) = 107 | AQMRPTModel(row.rid,row.mid,row.state,row.county,row.year,row.value,row.total,row.valid) 108 | val AQMRPTLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toAQMRPT _) 109 | val AQMRPTStream = AQMRPTLoader.fda_typedStream(AQMRPTQuery.result)(db)(256,256)()() 110 | 111 | def getIdsThenInsertAction: FDAUserTask[FDAROW] = row => { 112 | row match { 113 | case aqm: AQMRPTModel => 114 | if (aqm.valid) { 115 | val stateId = 0 //getStateID(aqm.state) 116 | val countyId = 0 //getCountyID(aqm.state,aqm.county) 117 | val action = NORMAQMQuery += NORMAQMModel(0,aqm.mid, stateId, countyId, aqm.year,aqm.value,aqm.total) 118 | fda_next(FDAActionRow(action)) 119 | } 120 | else fda_skip 121 | case _ => fda_skip 122 | } 123 | } 124 | val runner = FDAActionRunner(slick.jdbc.H2Profile) 125 | def runInsertAction: FDAUserTask[FDAROW] = row => 126 | row match { 127 | case FDAActionRow(action) => 128 | runner.fda_execAction(action)(db) 129 | fda_skip 130 | case _ => fda_skip 131 | } 132 | 133 | val cnt_start = System.currentTimeMillis() 134 | 135 | 136 | /* 137 | AQMRPTStream.take(100000) 138 | .appendTask(getIdsThenInsertAction) 139 | .appendTask(runInsertAction) 140 | .startRun 141 | //println(s"processing 10000 rows in a single thread in ${(System.currentTimeMillis - cnt_start)/1000} seconds") 142 | //processing 10000 rows in a single thread in 570 seconds 143 | //println(s"processing 20000 rows in a single thread in ${(System.currentTimeMillis - cnt_start)/1000} seconds") 144 | //processing 20000 rows in a single thread in 1090 seconds 145 | //println(s"processing 100000 rows in a single thread in ${(System.currentTimeMillis - cnt_start)/1000} seconds") 146 | //processing 100000 rows in a single thread in 2+ hrs 147 | */ 148 | 149 | implicit val strategy = Strategy.fromCachedDaemonPool("cachedPool") 150 | // implicit val strategy = Strategy.fromFixedDaemonPool(6) 151 | fda_runPar(AQMRPTStream.toPar(getIdsThenInsertAction))(4) 152 | .appendTask(runInsertAction) 153 | .startRun 154 | 155 | //println(s"processing 10000 rows parallelly in ${(System.currentTimeMillis - cnt_start)/1000} seconds") 156 | // processing 10000 rows parallelly in 316 seconds 157 | //println(s"processing 20000 rows parallelly in ${(System.currentTimeMillis - cnt_start)/1000} seconds") 158 | //processing 20000 rows parallelly in 614 seconds 159 | println(s"processing 100000 rows parallelly in ${(System.currentTimeMillis - cnt_start)/1000} seconds") 160 | //processing 100000 rows parallelly in 3885 seconds 161 | 162 | } 163 | -------------------------------------------------------------------------------- /src/main/scala/examples/ParallelExecution.scala: -------------------------------------------------------------------------------- 1 | package examples 2 | import slick.jdbc.meta._ 3 | import com.bayakala.funda._ 4 | import api._ 5 | import scala.language.implicitConversions 6 | import scala.concurrent.ExecutionContext.Implicits.global 7 | import scala.concurrent.duration._ 8 | import scala.concurrent.{Await, Future} 9 | import scala.util.{Failure, Success} 10 | import slick.jdbc.H2Profile.api._ 11 | import com.bayakala.funda.samples.SlickModels._ 12 | 13 | 14 | object ParallelExecution extends App { 15 | 16 | val db = Database.forConfig("h2db") 17 | 18 | //drop original table schema 19 | val futVectorTables = db.run(MTable.getTables) 20 | 21 | val futDropTable = futVectorTables.flatMap{ tables => { 22 | val tableNames = tables.map(t => t.name.name) 23 | if (tableNames.contains(NORMAQMQuery.baseTableRow.tableName)) 24 | db.run(NORMAQMQuery.schema.drop) 25 | else Future(():Unit) 26 | } 27 | }.andThen { 28 | case Success(_) => println(s"Table ${NORMAQMQuery.baseTableRow.tableName} dropped successfully! ") 29 | case Failure(e) => println(s"Failed to drop Table ${NORMAQMQuery.baseTableRow.tableName}, it may not exist! Error: ${e.getMessage}") 30 | } 31 | Await.ready(futDropTable,Duration.Inf) 32 | 33 | //create new table to refine AQMRawTable 34 | val actionCreateTable = NORMAQMQuery.schema.create 35 | val futCreateTable = db.run(actionCreateTable).andThen { 36 | case Success(_) => println("Table created successfully!") 37 | case Failure(e) => println(s"Table may exist already! Error: ${e.getMessage}") 38 | } 39 | //would carry on even fail to create table 40 | Await.ready(futCreateTable,Duration.Inf) 41 | 42 | 43 | //truncate data, only available in slick 3.2.1 44 | val futTruncateTable = futVectorTables.flatMap{ tables => { 45 | val tableNames = tables.map(t => t.name.name) 46 | if (tableNames.contains(NORMAQMQuery.baseTableRow.tableName)) 47 | db.run(NORMAQMQuery.schema.truncate) 48 | else Future(():Unit) 49 | } 50 | }.andThen { 51 | case Success(_) => println(s"Table ${NORMAQMQuery.baseTableRow.tableName} truncated successfully!") 52 | case Failure(e) => println(s"Failed to truncate Table ${NORMAQMQuery.baseTableRow.tableName}! Error: ${e.getMessage}") 53 | } 54 | Await.ready(futDropTable,Duration.Inf) 55 | 56 | //a conceived task for the purpose of resource consumption 57 | //getting id with corresponding name from STATES table 58 | def getStateID(state: String): Int = { 59 | //create a stream for state id with state name 60 | implicit def toState(row: StateTable#TableElementType) = StateModel(row.id,row.name) 61 | val stateLoader = FDAViewLoader(slick.jdbc.H2Profile)(toState _) 62 | val stateSeq = stateLoader.fda_typedRows(StateQuery.result)(db).toSeq 63 | //constructed a Stream[Task,String] 64 | val stateStream = fda_staticSource(stateSeq)() 65 | var id = -1 66 | def getid: FDAUserTask[FDAROW] = row => { 67 | row match { 68 | case StateModel(stid,stname) => //target row type 69 | if (stname.contains(state)) { 70 | id = stid 71 | fda_break //exit 72 | } 73 | else fda_skip //take next row 74 | case _ => fda_skip 75 | } 76 | } 77 | stateStream.appendTask(getid).startRun 78 | id 79 | } 80 | //another conceived task for the purpose of resource consumption 81 | //getting id with corresponding names from COUNTIES table 82 | def getCountyID(state: String, county: String): Int = { 83 | //create a stream for county id with state name and county name 84 | implicit def toCounty(row: CountyTable#TableElementType) = CountyModel(row.id,row.name) 85 | val countyLoader = FDAViewLoader(slick.jdbc.H2Profile)(toCounty _) 86 | val countySeq = countyLoader.fda_typedRows(CountyQuery.result)(db).toSeq 87 | //constructed a Stream[Task,String] 88 | val countyStream = fda_staticSource(countySeq)() 89 | var id = -1 90 | def getid: FDAUserTask[FDAROW] = row => { 91 | row match { 92 | case CountyModel(cid,cname) => //target row type 93 | if (cname.contains(state) && cname.contains(county)) { 94 | id = cid 95 | fda_break //exit 96 | } 97 | else fda_skip //take next row 98 | case _ => fda_skip 99 | } 100 | } 101 | countyStream.appendTask(getid).startRun 102 | id 103 | } 104 | 105 | //process input row and produce action row to insert into NORMAQM 106 | def getIdsThenInsertAction: FDAUserTask[FDAROW] = row => { 107 | row match { 108 | case aqm: AQMRPTModel => 109 | if (aqm.valid) { 110 | val stateId = getStateID(aqm.state) 111 | val countyId = getCountyID(aqm.state,aqm.county) 112 | val action = NORMAQMQuery += NORMAQMModel(0,aqm.mid, stateId, countyId, aqm.year,aqm.value,aqm.total) 113 | fda_next(FDAActionRow(action)) 114 | } 115 | else fda_skip 116 | case _ => fda_skip 117 | } 118 | } 119 | //runner for the action rows 120 | val runner = FDAActionRunner(slick.jdbc.H2Profile) 121 | def runInsertAction: FDAUserTask[FDAROW] = row => 122 | row match { 123 | case FDAActionRow(action) => 124 | runner.fda_execAction(action)(db) 125 | fda_skip 126 | case _ => fda_skip 127 | } 128 | 129 | //create parallel sources 130 | //get a stream of years 131 | val qryYears = AQMRPTQuery.map(_.year).distinct 132 | case class Years(year: Int) extends FDAROW 133 | 134 | implicit def toYears(y: Int) = Years(y) 135 | 136 | val yearViewLoader = FDAViewLoader(slick.jdbc.H2Profile)(toYears _) 137 | val yearSeq = yearViewLoader.fda_typedRows(qryYears.result)(db).toSeq 138 | val yearStream = fda_staticSource(yearSeq)() 139 | 140 | //strong row type 141 | implicit def toAQMRPT(row: AQMRPTTable#TableElementType) = 142 | AQMRPTModel(row.rid, row.mid, row.state, row.county, row.year, row.value, row.total, row.valid) 143 | 144 | //shared stream loader when operate in parallel mode 145 | val AQMRPTLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toAQMRPT _) 146 | 147 | //loading rows with year yr 148 | def loadRowsInYear(yr: Int) = { 149 | //a new query 150 | val query = AQMRPTQuery.filter(row => row.year === yr) 151 | //reuse same loader 152 | AQMRPTLoader.fda_typedStream(query.result)(db)(256, 256)(println(s"End of stream ${yr}!!!!!!"))() 153 | } 154 | 155 | //loading rows by year 156 | def loadRowsByYear: FDASourceLoader = row => { 157 | row match { 158 | case Years(y) => loadRowsInYear(y) //produce stream of the year 159 | case _ => fda_appendRow(FDANullRow) 160 | } 161 | } 162 | 163 | 164 | //start counter 165 | val cnt_start = System.currentTimeMillis() 166 | 167 | def showRecord: FDAUserTask[FDAROW] = row => { 168 | row match { 169 | case Years(y) => println(y); fda_skip 170 | case aqm: AQMRPTModel => 171 | println(s"${aqm.year} $aqm") 172 | fda_next(aqm) 173 | case FDAActionRow(action) => 174 | println(s"${action}") 175 | fda_skip 176 | case _ => fda_skip 177 | } 178 | } 179 | 180 | //the following is a process of composition of stream combinators 181 | //get parallel source constructor 182 | val parSource = yearStream.toParSource(loadRowsByYear) 183 | //produce a stream from parallel sources 184 | val source = fda_par_source(parSource)(4) 185 | //turn getIdsThenInsertAction into parallel task 186 | val parTasks = source.toPar(getIdsThenInsertAction) 187 | //runPar to produce a new stream 188 | val actionStream =fda_runPar(parTasks)(4) 189 | //turn runInsertAction into parallel task 190 | val parRun = actionStream.toPar(runInsertAction) 191 | //runPar and carry out by startRun 192 | fda_runPar(parRun)(2).startRun 193 | 194 | println(s"processing 219400 rows parallelly in ${(System.currentTimeMillis - cnt_start)/1000} seconds") 195 | 196 | 197 | 198 | } 199 | -------------------------------------------------------------------------------- /license.txt: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "{}" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright {yyyy} {name of copyright owner} 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/package.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala 2 | 3 | /** 4 | * FunDA core types, global imports and fs2 stream method injection 5 | */ 6 | package object funda { 7 | import fs2._ 8 | import slick.dbio._ 9 | import scala.concurrent.Future 10 | import akka.stream.stage._ 11 | 12 | implicit val fda_strategy = Strategy.fromFixedDaemonPool(4) 13 | implicit val fda_scheduler = Scheduler.fromFixedDaemonPool(4) 14 | 15 | /** fs2 manned halt type 16 | * terminateNow indicates to FDADataStream.pushData intent to stop 17 | * anytime now during enqueue process 18 | */ 19 | class Fs2Terminator { 20 | var terminateNow = false 21 | def reset = terminateNow = false 22 | def stopASAP = terminateNow = true 23 | } 24 | /** default killswitch for fs2 25 | * declare separate instances for multiple concurrent running stream control 26 | * by providing explicitly killSwitch parameter 27 | */ 28 | implicit object Fs2KillSwitch extends Fs2Terminator 29 | /** akka manned halt type 30 | * terminateNow indicates to FDADataStream.Fs2Gate intent to stop 31 | * anytime now during enqueue process 32 | */ 33 | class AkkaTerminator{ 34 | var callback: AsyncCallback[Unit] = null 35 | def stopASAP = { 36 | if (callback != null) { 37 | callback.invoke(()) 38 | } 39 | } 40 | } 41 | /** default killswitch for akka 42 | * declare separate instances for multiple concurrent running stream control 43 | * by providing explicitly killSwitch parameter 44 | */ 45 | implicit object AkkaKillSwitch extends AkkaTerminator 46 | 47 | /** 数据处理管道 48 | * a stream of data or action rows 49 | * @tparam ROW type of row 50 | */ 51 | type FDAPipeLine[ROW] = Stream[Task, ROW] 52 | 53 | /** 数据作业节点 54 | * a work node appended to stream to perform user action 55 | * @tparam ROW type of row 56 | */ 57 | type FDAWorkNode[ROW] = Pipe[Task, ROW, ROW] 58 | 59 | /** 数据管道开关阀门，从此处获得管道内数据 60 | * a handle to get rows from upstream 61 | * @tparam ROW type of row 62 | */ 63 | type FDAValve[ROW] = Handle[Task, ROW] 64 | 65 | /** 管道连接器 66 | * gate to send rows downstream 67 | * @tparam ROW type of row 68 | */ 69 | type FDAPipeJoint[ROW] = Pull[Task, ROW, Unit] 70 | 71 | /** 作业类型 72 | * user define function to be performed at a FDAWorkNode 73 | * given a row from upstream, return Option[List[ROW]] as follows: 74 | * fda_skip -> Some(Nil) : skip sending the current row 75 | * fda_next -> Some(List(r1,r2...)): send r1,r2... downstream 76 | * fda_break -> None : halt stream, end of process 77 | * @tparam ROW type of row 78 | */ 79 | type FDAUserTask[ROW] = (ROW) => (Option[List[ROW]]) 80 | 81 | /** 数据源构建器类型 82 | * a function type to produce a stream from input row 83 | * used to be turn into FDAParSource by toParSource 84 | */ 85 | type FDASourceLoader = FDAROW => FDAPipeLine[FDAROW] 86 | 87 | /** 合计作业类型 88 | * user define function with aggregation effect to be performed at a FDAWorkNode 89 | * given current aggregation value and row from upstream, 90 | * return updated aggregation value and Option[List[ROW]] as follows: 91 | * fda_skip -> Some(Nil) : skip sending the current row 92 | * fda_next -> Some(List(r1,r2...)): send r1,r2... downstream 93 | * fda_break -> None : halt stream, end of process 94 | * @tparam AGGR type of aggregation 95 | * @tparam ROW type of row 96 | */ 97 | type FDAAggrTask[AGGR,ROW] = (AGGR,ROW) => (AGGR,Option[List[ROW]]) 98 | 99 | /** 并行作业类型 100 | * stream of streams type for parallel running user action 101 | * use stream.toPar to convert from FDAUserTask 102 | */ 103 | type FDAParTask = Stream[Task,Stream[Task,Option[List[FDAROW]]]] 104 | 105 | /** 并行数据源类型 106 | * source of sources type for parallel loading data sources 107 | * use stream.toParSource to convert from FDASourceLoader 108 | */ 109 | type FDAParSource = Stream[Task,Stream[Task,FDAROW]] 110 | 111 | /** 数据行类型 112 | * topmost generic row type 113 | */ 114 | trait FDAROW 115 | 116 | /** 117 | * a EOS object used to signify end of stream 118 | */ 119 | case object FDANullRow extends FDAROW 120 | 121 | /** 122 | * capture exception in a row 123 | * @param e 124 | */ 125 | case class FDAErrorRow(e: Exception) extends FDAROW 126 | 127 | /** 128 | * manually emit a row such as FDANullRow or FDAErrorRow 129 | * @example {{{ 130 | * //loading rows by year 131 | * def loadRowsByYear: FDASourceLoader = row => { 132 | * row match { 133 | * case Years(y) => loadRowsInYear(y) //produce stream of the year 134 | * case _ => fda_appendRow(FDANullRow) 135 | * } 136 | * } 137 | * }}} 138 | * @param row row to emit 139 | * @return new stream 140 | */ 141 | def fda_appendRow(row: FDAROW): FDAPipeLine[FDAROW] = Stream(row) 142 | 143 | /** 144 | * runnable action type 145 | */ 146 | type FDAAction = DBIO[Int] 147 | 148 | /** 149 | * action row type. can have futher distinct child type as fullows: 150 | * @example {{{ 151 | * scala> class MyActionRow(action: FDAAction) extends FDAActionRow(action) 152 | * }}} 153 | * @param action runnable action 154 | */ 155 | case class FDAActionRow(action: FDAAction) extends FDAROW 156 | 157 | /** 158 | * methods injected to fs2Stream 159 | */ 160 | implicit class toFDAOps(fs2Stream: FDAPipeLine[FDAROW]) { 161 | /** 162 | * append a user task t to stream 163 | * @example {{{ 164 | * val streamAllTasks = streamAQMRaw.appendTask(filterRows) 165 | * .appendTask(toAction) 166 | * .appendTask(runActionRow) 167 | * }}} 168 | * @param t user defined function 169 | * @return new stream 170 | */ 171 | def appendTask(t: FDAUserTask[FDAROW]): FDAPipeLine[FDAROW] = 172 | fs2Stream.through(FDATask.fda_execUserTask(t)) 173 | 174 | /** 175 | * append a user defined aggregation task t 176 | * @example {{{ 177 | * //user defined aggregator type. 178 | * case class Accu(state: String, county: String, year: Int, count: Int, sumOfValue: Int) 179 | * 180 | * aqmrStream.aggregateTask(Accu("","",0,0,0),aggregateValue) 181 | * .appendTask(toAction) 182 | * .appendTask(runActionRow) 183 | * .startRun 184 | * }}} 185 | * @param aggr initial value of aggregation 186 | * @param t user defined task 187 | * @tparam AGGR type of aggr 188 | * @return new stream 189 | */ 190 | def aggregateTask[AGGR](aggr: AGGR, t: FDAAggrTask[AGGR, FDAROW]): FDAPipeLine[FDAROW] = 191 | fs2Stream.through(FDATask.fda_aggregate(aggr, t)) 192 | 193 | /** 194 | * replace stream[Task,ROW].run.unsafeRun 195 | * @example {{{ 196 | * streamAQMRaw.appendTask(filterRows) 197 | * .appendTask(toAction) 198 | * .appendTask(runActionRow) 199 | * .startRun 200 | * }}} 201 | */ 202 | def startRun: Unit = fs2Stream.run.unsafeRun 203 | 204 | /** 205 | * replace stream[Task,ROW].run.unsafeRunAsyncFuture 206 | * returns immediately 207 | * 208 | * @return Future 209 | */ 210 | def startFuture[A]: Future[Unit] = fs2Stream.run.unsafeRunAsyncFuture 211 | 212 | 213 | /** 214 | * turn user task into type for parallel computation 215 | * @example {{{ 216 | * //runner for the action rows 217 | * val runner = FDAActionRunner(slick.jdbc.H2Profile) 218 | * def runInsertAction: FDAUserTask[FDAROW] = row => 219 | * row match { 220 | * case FDAActionRow(action) => 221 | * runner.fda_execAction(action)(db) 222 | * fda_skip 223 | * case _ => fda_skip 224 | * } 225 | * 226 | * //turn runInsertAction into parallel task 227 | * val parRun = actionStream.toPar(runInsertAction) 228 | * }}} 229 | * @param st user defined task 230 | * @return stream of streams 231 | */ 232 | def toPar(st: FDAUserTask[FDAROW]): FDAParTask = 233 | fs2Stream.map { row => 234 | Stream.eval(Task { 235 | st(row) 236 | }) 237 | } 238 | 239 | /** 240 | * turn a single stream into parallel sources 241 | * @example {{{ 242 | * //loading rows with year yr 243 | * def loadRowsInYear(yr: Int) = { 244 | * //a new query 245 | * val query = AQMRPTQuery.filter(row => row.year === yr) 246 | * //reuse same loader 247 | * AQMRPTLoader.fda_typedStream(query.result)(db)(256, 256)(println(s"End of stream ${yr}!!!!!!")) 248 | * } 249 | * 250 | * //loading rows by year 251 | * def loadRowsByYear: FDASourceLoader = row => { 252 | * row match { 253 | * case Years(y) => loadRowsInYear(y) //produce stream of the year 254 | * case _ => fda_appendRow(FDANullRow) 255 | * } 256 | * } 257 | * 258 | * //produce a stream from parallel sources 259 | * val source = fda_par_source(parSource)(4) 260 | * }}} 261 | * @param load stream constructing function: FDAROW => FDAPipeLine[FDAROW] 262 | * @return stream of streams 263 | */ 264 | def toParSource(load: FDASourceLoader): FDAParSource = 265 | fs2Stream.map(row => load(row)) 266 | 267 | } 268 | 269 | 270 | /** methods to run an user defined function on FDAPipeLine*/ 271 | object FDATask { //作业节点工作方法 272 | /** 273 | * returns state of next worknode. using fs2 Handle of Pull object, 274 | * take the next element and apply function task and determine new state of stream 275 | * @param task user defined function: ROW => Option[List[ROW]] 276 | * returns an Option[List[ROW]]] value signifying movement downstream 277 | * as follows: 278 | * Some(Nil) : skip sending the current row 279 | * Some(List(r1,r2...)): send r1,r2... downstream 280 | * None : halt stream, end of process 281 | * @tparam ROW row type: FDAROW or FDAActionROW 282 | * @return new state of stream 283 | */ 284 | private[funda] def fda_execUserTask[ROW](task: FDAUserTask[ROW]): FDAWorkNode[ROW] = { 285 | def go: FDAValve[ROW] => FDAPipeJoint[ROW] = h => { 286 | h.receive1Option { 287 | case Some((r, h)) => task(r) match { 288 | case Some(lx) => lx match { 289 | case Nil => go(h) 290 | case _ => Pull.output(Chunk.seq(lx)) >> go(h) 291 | } 292 | case None => task(FDANullRow.asInstanceOf[ROW]) match { 293 | case Some(lx) => lx match { 294 | case Nil => Pull.done 295 | case _ => Pull.output(Chunk.seq(lx)) >> Pull.done 296 | } 297 | case _ => Pull.done 298 | } 299 | } 300 | case None => task(FDANullRow.asInstanceOf[ROW]) match { 301 | case Some(lx) => lx match { 302 | case Nil => Pull.done 303 | case _ => Pull.output(Chunk.seq(lx)) >> Pull.done 304 | } 305 | case _ => Pull.done 306 | } 307 | } 308 | } 309 | in => in.pull(go) 310 | } 311 | /** 312 | * returns state of next worknode and some aggregation defined inside user function. 313 | * execute user defined function with internal aggregation mechanism by means of 314 | * functional state transition style of passing in state and return new state. 315 | * take in current aggregation and next row, apply user function on both 316 | * and determine new state of stream 317 | * @param aggr user selected type of aggregation such as Int, (Int,Int) ... 318 | * @param task user defined function: (AGGR,ROW) => (AGGR,Option[List[ROW]]) 319 | * take in current aggregation and row, 320 | * and return new aggregation and Option[List[ROW]] with meaning of: 321 | * Some(Nil) : skip sending the current row 322 | * Some(List(r1,r2...)): send r1,r2... downstream 323 | * None : halt stream, end of process 324 | * @tparam AGGR type of aggr 325 | * @tparam ROW type of row 326 | * @return new state of stream 327 | */ 328 | private[funda] def fda_aggregate[AGGR,ROW](aggr: AGGR, task: FDAAggrTask[AGGR,ROW]): FDAWorkNode[ROW] = { 329 | def go(acc: AGGR): FDAValve[ROW] => FDAPipeJoint[ROW] = h => { 330 | h.receive1Option { 331 | case Some((r, h)) => task(acc,r) match { 332 | case (a,Some(lx)) => lx match { 333 | case Nil => go(a)(h) 334 | case _ => Pull.output(Chunk.seq(lx)) >> go(a)(h) 335 | } 336 | case (a,None) => task(a,FDANullRow.asInstanceOf[ROW]) match { 337 | case (a,Some(lx)) => lx match { 338 | case Nil => Pull.done 339 | case _ => Pull.output(Chunk.seq(lx)) >> Pull.done 340 | } 341 | case _ => Pull.done 342 | } 343 | } 344 | case None => task(acc,FDANullRow.asInstanceOf[ROW]) match { 345 | case (a,Some(lx)) => lx match { 346 | case Nil => Pull.done 347 | case _ => Pull.output(Chunk.seq(lx)) >> Pull.done 348 | } 349 | case _ => Pull.done 350 | } 351 | } 352 | } 353 | in => in.pull(go(aggr)) 354 | } 355 | 356 | 357 | } 358 | 359 | 360 | } 361 | 362 | -------------------------------------------------------------------------------- /src/main/scala/com/bayakala/funda/fdasources/FDADataStream.scala: -------------------------------------------------------------------------------- 1 | package com.bayakala.funda.fdasources 2 | 3 | import fs2._ 4 | import play.api.libs.iteratee._ 5 | import com.bayakala.funda._ 6 | import slick.jdbc.JdbcProfile 7 | 8 | import akka.actor._ 9 | import akka.stream.scaladsl._ 10 | import akka.stream._ 11 | import akka.stream.stage._ 12 | import akka.stream.stage.{GraphStage, GraphStageLogic} 13 | 14 | /** stream loader class wrapper */ 15 | trait FDADataStream { 16 | 17 | /** running Slick DBIOAction to produce a data stream conforming to reactive-streams api. 18 | * provide strong typed result conversion if required 19 | * @param slickProfile Slick jdbc profile such as 'slick.jdbc.H2Profile' 20 | * @param convert a defined implicit type conversion function. 21 | * from SOURCE type to TARGET type, set to null if not required 22 | * @tparam SOURCE source type, result type of DBIOAction, most likely a tuple type 23 | * @tparam TARGET final converted type, most likely a case class type 24 | */ 25 | class FDAStreamLoader[SOURCE, TARGET](slickProfile: JdbcProfile, convert: SOURCE => TARGET) { 26 | 27 | import slickProfile.api._ 28 | 29 | /** 30 | * returns a reactive-stream from Slick DBIOAction result 31 | * using play-iteratees and fs2 queque to connect to slick data stream publisher 32 | * provide facade for error handler and finalizer to support exception and cleanup handling 33 | * also provide stream element conversion from SOURCE type to TARGET type 34 | * @example {{{ 35 | * val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toTypedRow _) 36 | * val streamSource = streamLoader.fda_typedStream(aqmQuery.result)(db)(512,16,100)()() 37 | * val safeStreamSource = streamLoader.fda_typedStream(aqmQuery.result)(db)(512,16)( 38 | * println("the end finally!"))(killSwitch) 39 | * }}} 40 | * @param action a Slick DBIOAction to produce query results 41 | * @param slickDB Slick database object 42 | * @param fetchSize number of rows cached during database read 43 | * @param queSize size of queque used by iteratee as cache to pass elements to fs2 stream 44 | * @param take take first 'take' elements 45 | * @param finalizer cleanup callback 46 | * @param killSwitch use killSwitch.stopASAP to halt stream 47 | * @param convert just a measure to guarantee conversion function is defined 48 | * when this function is used there has to be a converter defined 49 | * implicitly in compile time 50 | * @return a reactive-stream of TARGET row type elements 51 | */ 52 | def fda_typedStream(action: DBIOAction[Iterable[SOURCE],Streaming[SOURCE],Effect.Read])( 53 | slickDB: Database)( 54 | fetchSize: Int, queSize: Int, take: Int = 0)( 55 | finalizer: => Unit = ())( 56 | killSwitch: Fs2Terminator = Fs2KillSwitch)( 57 | implicit convert: SOURCE => TARGET) 58 | : FDAPipeLine[TARGET] = { 59 | val disableAutocommit = SimpleDBIO(_.connection.setAutoCommit(false)) 60 | val action_ = action.withStatementParameters(fetchSize = fetchSize) 61 | val publisher = slickDB.stream(disableAutocommit andThen action_) 62 | val enumerator = streams.IterateeStreams.publisherToEnumerator(publisher) 63 | 64 | val s = Stream.eval(async.boundedQueue[Task,Option[SOURCE]](queSize)).flatMap { q => 65 | Task { Iteratee.flatten(enumerator |>> pushData(killSwitch,take,q)).run }.unsafeRunAsyncFuture() 66 | pipe.unNoneTerminate(q.dequeue).map {row => convert(row)} 67 | } 68 | s.onFinalize(Task.delay(finalizer)) 69 | 70 | } 71 | /** 72 | * returns a reactive-stream from Slick DBIOAction result 73 | * using akka-stream to connect to slick data stream publisher 74 | * provide facade for error handler and finalizer to support exception and cleanup handling 75 | * @example {{{ 76 | * val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toTypedRow _) 77 | * val streamSource = streamLoader.fda_akkaTypedStream(aqmQuery.result)(db)(512,2,100)()() 78 | * val safeStreamSource = streamLoader.fda_akkaTypedStream(aqmQuery.result)(db)(512,2)( 79 | * println("the end finally!"))(killSwitch) 80 | * }}} 81 | * @param action a Slick DBIOAction to produce query results 82 | * @param slickDB Slick database object 83 | * @param fetchSize number of rows cached during database read 84 | * @param queSize size of queque used akka-stream as cache to pass elements to fs2 queue 85 | * @param take take first 'take' elements 86 | * @param finalizer cleanup callback 87 | * @param killSwitch use killSwitch.stopASAP to halt stream 88 | * @param convert just a measure to guarantee conversion function is defined 89 | * when this function is used there has to be a converter defined 90 | * implicitly in compile time 91 | * @return a reactive-stream of SOURCE row type elements 92 | */ 93 | def fda_akkaTypedStream(action: DBIOAction[Iterable[SOURCE],Streaming[SOURCE],Effect.Read])( 94 | slickDB: Database)( 95 | fetchSize: Int, queSize: Int, take: Int = 0)( 96 | finalizer: => Unit = ())( 97 | killSwitch: AkkaTerminator = AkkaKillSwitch)( 98 | implicit convert: SOURCE => TARGET) 99 | : FDAPipeLine[TARGET] = { 100 | val disableAutocommit = SimpleDBIO(_.connection.setAutoCommit(false)) 101 | val action_ = action.withStatementParameters(fetchSize = fetchSize) 102 | val publisher = slickDB.stream(disableAutocommit andThen action_) 103 | implicit val actorSys = ActorSystem("actor-system") 104 | implicit val ec = actorSys.dispatcher 105 | implicit val mat = ActorMaterializer() 106 | // construct akka source 107 | val akkaSource = Source.fromPublisher[SOURCE](publisher) 108 | 109 | val s = Stream.eval(async.boundedQueue[Task,Option[SOURCE]](queSize)) 110 | .flatMap { q => 111 | Task(akkaSource.to(new FS2Gate[SOURCE](killSwitch, take, q)).run).unsafeRunAsyncFuture //enqueue Task(new thread) 112 | pipe.unNoneTerminate(q.dequeue).map {row => convert(row)} //dequeue in current thread 113 | } 114 | s.onFinalize{Task.delay{actorSys.terminate();finalizer}} 115 | } 116 | /** 117 | * returns a reactive-stream from Slick DBIOAction result 118 | * using play-iteratees and fs2 queque to connect to slick data stream publisher 119 | * provide facade for error handler and finalizer to support exception and cleanup handling 120 | * @example {{{ 121 | * val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)() 122 | * val streamSource = streamLoader.fda_plainStream(aqmQuery.result)(db)(512,16, 100)()() 123 | * val safeStreamSource = streamLoader.fda_plainStream(aqmQuery.result)(db)(512,16)( 124 | * println("the end finally!"))(killSwitch) 125 | * }}} 126 | * @param action a Slick DBIOAction to produce query results 127 | * @param slickDB Slick database object 128 | * @param fetchSize number of rows cached during database read 129 | * @param queSize size of queque used by iteratee as cache to pass elements to fs2 stream 130 | * @param take take first 'take' elements 131 | * @param finalizer cleanup callback 132 | * @param killSwitch use killSwitch.stopASAP to halt stream 133 | * @return a reactive-stream of SOURCE row type elements 134 | */ 135 | def fda_plainStream(action: DBIOAction[Iterable[SOURCE],Streaming[SOURCE],Effect.Read])( 136 | slickDB: Database)( 137 | fetchSize: Int, queSize: Int, take: Int = 0)( 138 | finalizer: => Unit = ())( 139 | implicit killSwitch: Fs2Terminator): FDAPipeLine[SOURCE] = { 140 | val disableAutocommit = SimpleDBIO(_.connection.setAutoCommit(false)) 141 | val action_ = action.withStatementParameters(fetchSize = fetchSize) 142 | val publisher = slickDB.stream(disableAutocommit andThen action_) 143 | val enumerator = streams.IterateeStreams.publisherToEnumerator(publisher) 144 | 145 | val s = Stream.eval(async.boundedQueue[Task,Option[SOURCE]](queSize)).flatMap { q => 146 | Task { Iteratee.flatten(enumerator |>> pushData(killSwitch,take,q)).run }.unsafeRunAsyncFuture() 147 | pipe.unNoneTerminate(q.dequeue) 148 | } 149 | s.onFinalize(Task.delay(finalizer)) 150 | } 151 | /** 152 | * returns a reactive-stream from Slick DBIOAction result 153 | * using akka-stream to connect to slick data stream publisher 154 | * provide facade for error handler and finalizer to support exception and cleanup handling 155 | * @example {{{ 156 | * val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)() 157 | * val streamSource = streamLoader.fda_akkaPlainStream(aqmQuery.result)(db)(512,2,100)()() 158 | * val safeStreamSource = streamLoader.fda_plainStream(aqmQuery.result)(db)(512,2)( 159 | * println("the end finally!"))(killSwitch) 160 | * }}} 161 | * @param action a Slick DBIOAction to produce query results 162 | * @param slickDB Slick database object 163 | * @param fetchSize number of rows cached during database read 164 | * @param queSize size of queque used akka-stream as cache to pass elements to fs2 queue 165 | * @param take take first 'take' elements 166 | * @param finalizer cleanup callback 167 | * @param killSwitch use killSwitch.stopASAP to halt stream 168 | * @return a reactive-stream of SOURCE row type elements 169 | */ 170 | def fda_akkaPlainStream(action: DBIOAction[Iterable[SOURCE],Streaming[SOURCE],Effect.Read])( 171 | slickDB: Database)( 172 | fetchSize: Int, queSize: Int, take: Int = 0)( 173 | finalizer: => Unit = ())( 174 | implicit killSwitch: AkkaTerminator): FDAPipeLine[SOURCE] = { 175 | val disableAutocommit = SimpleDBIO(_.connection.setAutoCommit(false)) 176 | val action_ = action.withStatementParameters(fetchSize = fetchSize) 177 | val publisher = slickDB.stream(disableAutocommit andThen action_) 178 | implicit val actorSys = ActorSystem("actor-system") 179 | implicit val ec = actorSys.dispatcher 180 | implicit val mat = ActorMaterializer() 181 | // construct akka source 182 | val akkaSource = Source.fromPublisher[SOURCE](publisher) 183 | 184 | val s = Stream.eval(async.boundedQueue[Task,Option[SOURCE]](queSize)) 185 | .flatMap { q => 186 | Task(akkaSource.to(new FS2Gate[SOURCE](killSwitch, take, q)).run).unsafeRunAsyncFuture //enqueue Task(new thread) 187 | pipe.unNoneTerminate(q.dequeue) //dequeue in current thread 188 | } 189 | s.onFinalize{Task.delay{actorSys.terminate();finalizer}} 190 | } 191 | 192 | /** 193 | * consume input from enumerator by pushing each element into q queque 194 | * end and produce error when enqueque could not be completed in timeout 195 | * @tparam R stream element type 196 | * @param killSwitch object with killSwitch.stopASAP to halt stream 197 | * @param take emit the first 'take' elements 198 | * @param q queque for cache purpose 199 | * @return iteratee in new state 200 | */ 201 | private def pushData[R](killSwitch: Fs2Terminator, take: Int, q: async.mutable.Queue[Task,Option[R]]): Iteratee[R,Unit] = Cont { 202 | case Input.EOF => 203 | q.enqueue1(None).unsafeRun 204 | Done((), Input.Empty) 205 | case Input.Empty => pushData(killSwitch,take,q) 206 | case Input.El(e) => 207 | if (take >= 0 && !killSwitch.terminateNow) { 208 | q.enqueue1(Some(e)).unsafeRun 209 | pushData(killSwitch, if(take == 0) 0 else {if (take == 1) -1 else take - 1}, q) 210 | } 211 | else { 212 | killSwitch.reset 213 | q.enqueue1(None).unsafeRun 214 | Done((), Input.Empty) 215 | } 216 | } 217 | /** 218 | * an akka-stream graph stage that connects akka-stream-source to fs2 flow through 219 | * ffs2.async.mutable.Queue structure 220 | * acts as a de-backpressurer to adjust emit rate to pull-model stream fs2 221 | * also takes care of manual halt and limiting first batch of element emission 222 | * @tparam T stream element type 223 | * @param killSwitch object with killSwitch.stopASAP to halt stream 224 | * @param take emit the first 'take' elements 225 | * @param q queque for cache purpose 226 | * @return iteratee in new state 227 | */ 228 | private class FS2Gate[T](killSwitch: AkkaTerminator, take: Int, q: fs2.async.mutable.Queue[Task,Option[T]]) extends GraphStage[SinkShape[T]] { 229 | val in = Inlet[T]("inport") 230 | val shape = SinkShape.of(in) 231 | 232 | override def createLogic(inheritedAttributes: Attributes): GraphStageLogic = 233 | new GraphStageLogic(shape) with InHandler { 234 | override def preStart(): Unit = { 235 | if (killSwitch != null) { 236 | val callback = getAsyncCallback[Unit] { (_) => 237 | killStream = true 238 | } 239 | killSwitch.callback = callback 240 | } 241 | pull(in) //initiate stream elements movement 242 | super.preStart() 243 | } 244 | var take_ = take 245 | var killStream = false 246 | override def onPush(): Unit = { 247 | if (killStream) take_ = -1 248 | q.enqueue1{ 249 | if ( take_ >= 0 ) 250 | Some(grab(in)) 251 | else 252 | None 253 | }.unsafeRun() 254 | pull(in) 255 | if ( take_ < 0) completeStage() 256 | if (take_ == 1) 257 | take_ = -1 258 | else 259 | if (take_ != 0) take_ -= 1 260 | } 261 | 262 | override def onUpstreamFinish(): Unit = { 263 | q.enqueue1(None).unsafeRun() 264 | completeStage() 265 | } 266 | 267 | override def onUpstreamFailure(ex: Throwable): Unit = { 268 | q.enqueue1(None).unsafeRun() 269 | completeStage() 270 | } 271 | 272 | setHandler(in,this) 273 | 274 | } 275 | } 276 | 277 | } 278 | 279 | /** 280 | * constructing FDAStreamLoader given slickProfile and converter 281 | */ 282 | object FDAStreamLoader { 283 | /** 284 | * constructor for FDAStreamLoader 285 | * @example {{{ 286 | * val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toTypedRow _) 287 | * val untypedLoader = FDAStreamLoader(slick.jdbc.H2Profile)() 288 | * }}} 289 | * @param slickProfile Slick jdbcprofile such as 'slick.jdbc.H2Profile' 290 | * @param converter a defined implicit type conversion function. 291 | * from SOURCE type to TARGET type, set to null if not required 292 | * @tparam SOURCE source type, result type of DBIOAction, most likely a tuple type 293 | * @tparam TARGET final converted type, most likely a case class type 294 | * @return a new FDAStreamLoader object 295 | */ 296 | def apply[SOURCE, TARGET](slickProfile: JdbcProfile)(converter: SOURCE => TARGET = null): FDAStreamLoader[SOURCE, TARGET] = 297 | new FDAStreamLoader[SOURCE, TARGET](slickProfile, converter) 298 | } 299 | } 300 | 301 | 302 | /** 303 | * for global imports 304 | */ 305 | object FDADataStream extends FDADataStream -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # FunDA (Functional Data Access) 2 | 3 | *FunDA* is a functional database access library designed to supplement *FRM* (*Functional-Relational-Mapper*) tools like *Slick*. 4 | While *FRM* tools bring type safe *language-integrated-query* and flexible query composition as well as powerful functional programming paradigm to it's users, the main focus is on data object access and thus in ways short on strength for data persistence support such as data row traversal operations that are so common in *ORM*. This short-fall plus a brand new functional programming style also make many *OOP* programmers from the *ORM* world using *FRM* quite uncomfortable or even somewhat awkward. On top of bringing back the familiar recordset operations to support data row processing, *FunDA* also adds in explicit parallel data processing capabilities as well as a simple user-defined programming model to lower requirements for functional programming skills such that with little getting-used-to a traditional *OOP* programmer could handle *FRM* so to make *FunDA* as a much practically productive tool. 5 | The core of *FunDA* is implemented with *scalaz-streams-fs2*. *FunDA* can be depicted as a workflow pipe with a sequence of work-nodes where user-defined data processing tasks could be plugged in. *FunDA* is implemented as a forward-only stream of rows representing pure data or query actions. User-defined-tasks at a work-node can intercept rows and run some processes with the context of each row. And these user-defined-tasks can be run parallelly through *FunDA*'s Parallelism. 6 | A typical *FunDA* program consists of a **source** and many **user-defined-tasks** as follows: 7 | 8 | ``` 9 | val streamSource = streamLoader.fda_typedStream(albumsInfo.result)(db)(512, 128)() 10 | 11 | streamSource.appendTask(transformData).appendTask(runActionQuery).appendTask(showResults).startRun 12 | ``` 13 | where "streamSource" is a *FunDA* stream **source** produced by loading data from database, and "transformData", "runActionQuery" and "showResults" are all user-defined-tasks each responsible to achieve some minimal distinctive effect. As the unique flavor of functional programming, these are functional combinators and can be composed in a specific order to perform a much bigger and complexed task. From the semantics of the *FunDA* program above we can make a wild guess that "transformData" would transform each data row to query actions and these query actions are executed by "runActionQuery" at the next work-node. 14 | #### how to use 15 | 16 | *FunDA* artifacts are currently published on Bintray. Add following in your build.sbt: 17 | 18 | ``` 19 | resolvers += Resolver.bintrayRepo("bayakala","maven") 20 | libraryDependencies += "com.bayakala" %% "funda" % "1.0.0-RC-01" withSources() withJavadoc() 21 | 22 | 23 | ``` 24 | for your info, *FunDA* allready includes the following dependencies: 25 | 26 | ``` 27 | libraryDependencies ++= Seq( 28 | "com.typesafe.slick" %% "slick" % "3.2.0", 29 | "com.h2database" % "h2" % "1.4.191", 30 | "com.typesafe.slick" %% "slick-hikaricp" % "3.2.0", 31 | "ch.qos.logback" % "logback-classic" % "1.1.7", 32 | "co.fs2" %% "fs2-core" % "0.9.4", 33 | "co.fs2" %% "fs2-io" % "0.9.4", 34 | "com.typesafe.play" % "play-iteratees-reactive-streams_2.11" % "2.6.0" 35 | ) 36 | 37 | ``` 38 | **Remarks:** users should set up their own *Slick* database configuration file application.conf in the resources directory. 39 | #### to run the examples 40 | There is a sample applicaion "funda-demo" located on github here: [www.github.com/bayakala/funda-demo](http://www.github.com/bayakala/funda-demo/) . It includes sample data located under resources/testdata/ and it is a bare cvs file. Import this file in your database before you run the examples. The examples should be run in the following order: 41 | 42 | ``` 43 | 1. StrongTypedRows.scala 44 | 2. UserDefinedTasks.scala 45 | 3. ParallelLoading.scala 46 | 4. ParallelTasks.scala 47 | 5. ParallelExecution.scala 48 | 6. ExceptionsAndFinalizers.scala 49 | ``` 50 | *download and try it. good luck and have fun!* 51 | ## The Principles 52 | 53 | *FunDA*'s workflow *FDAPipeLine* is a *scalaz-streams-fs2* and therefore is a *free-monad*. It is highly composable: 54 | 55 | ``` 56 | val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toTypedRow _) 57 | val source = streamLoader.fda_typedStream(aqmQuery.result)(db)(512,512)() 58 | val stream = source.filter{r => r.year > "1999"}.take(3).appendTask(showRecord) 59 | 60 | stream.startRun 61 | ``` 62 | as demonstrated above, we can compose stream anyway we want before **startRun** 63 | ### FunDA stream (The program) 64 | ##### strong-typed rows 65 | As mentioned above, FunDA programs are just composition of a **source** and a string of **user-defined-tasks** as a stream with data produced by **source** as rows. To facilitate stream operations we must convert data loaded from database into strong-typed rows. A practical case is that *Slick* usually returns query results in a collection of tuples. Thus we must take an extra step to convert them into user defined strong-typed case classes. 66 | The following code snippet demonstrates such conversion: 67 | 68 | ``` 69 | // aqmQuery.result returns Seq[(String,String,String,String)] 70 | val aqmQuery = aqmraw.map {r => (r.year,r.state,r.county,r.value)} 71 | // user designed strong typed resultset type. must extend FDAROW 72 | case class TypedRow(year: String, state: String, county: String, value: String) extends FDAROW 73 | // strong typed resultset conversion function. declared implicit to remind during compilation 74 | implicit def toTypedRow(row: (String,String,String,String)): TypedRow = 75 | TypedRow(row._1,row._2,row._3,row._4) 76 | 77 | // loader to read from database and convert result collection to strong typed collection 78 | val viewLoader = FDAViewLoader(slick.jdbc.H2Profile)(toTypedRow _) 79 | val dataSeq = viewLoader.fda_typedRows(aqmQuery.result)(db).toSeq 80 | // turn Seq collection into FunDA stream with strong-typed rows 81 | val aqmStream: FDAPipeLine[TypedRow] = fda_staticSource(dataSeq)() 82 | 83 | ``` 84 | ##### static view and dynamic streaming sources 85 | Static sources or views are data structures completely loaded into memory after returning from running a query. Stream sources are data streams as returned query results that are *reactive-streams* conformant. In other words stream sources are backend cached and motivated by back-pressure. *FunDA* provides functions to produce sources. The following is a demonstration of static view producing: 86 | 87 | ``` 88 | // loader to read from database and convert result collection to strong typed collection 89 | val viewLoader = FDAViewLoader(slick.jdbc.H2Profile)(toTypedRow _) 90 | val dataSeq = viewLoader.fda_typedRows(aqmQuery.result)(db).toSeq 91 | // turn Seq collection into FunDA stream with strong-typed rows 92 | val aqmView: FDAPipeLine[TypedRow] = fda_staticSource(dataSeq)() 93 | 94 | ``` 95 | stream source can be constructed as follows: 96 | 97 | ``` 98 | // strong typed source is also possible with Slick data streaming 99 | val streamLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toTypedRow _) 100 | val aqmStream: FDAPipeLine[TypedRow] = streamLoader.fda_typedStream(aqmQuery.result)(db)(512,512)() 101 | ``` 102 | as demonstrated above, both static collections and dynamic data streams can be transform into strong-typed-row sources. 103 | 104 | ### Control data flow 105 | 106 | Flow of rows in *FunDA* streams are controlled inside user-defined-tasks, in which a row is received from upstream and zero or one or more rows could be passed downstream. This means additional new rows could be constructed instantly and passed downstream inside these user-defined-tasks which makes *FunDA* logics much more flexible and powerful. Passing no row in a receive-send-loop is represented by skip. User could also halt stream by passing an end-of-stream signal downstream inside these user-defined-tasks. The following are some code samples: 107 | 108 | ``` 109 | user-defined-task type is defined as follows: 110 | type FDAUserTask[ROW] = (ROW) => (Option[List[ROW]]) 111 | /* user define function to be performed at a FDAWorkNode 112 | * given a row from upstream, return Option[List[ROW]] as follows: 113 | * fda_skip -> Some(Nil) : skip sending the current row 114 | * fda_next -> Some(List(r1,r2...)): send r1,r2... downstream 115 | * fda_break -> None : halt stream, end of process 116 | * @tparam ROW type of row 117 | */ 118 | 119 | // a example of user-defined-task 120 | def dancing: FDAUserTask[FDAROW] = row => { 121 | row match { 122 | case qmr: TypedRow => 123 | qmr.year match { 124 | case a if a < 1960 => 125 | // pass downstream untouched 126 | fda_next(qmr) 127 | case b if b < 1970 => 128 | // transform row 129 | fda_next(qmr.copy(year = 1970)) 130 | case c if c < 1980 => 131 | // pass alone with a new row. TypedRow is a case class 132 | fda_next(List(qmr,TypedRow(qmr.stare,"countyQQ",0.3,1980))) 133 | case d if d < 2000 => 134 | // do not process this row 135 | fda_skip 136 | case _ => 137 | // stop stream 138 | fda_break 139 | } 140 | // encounter unknown row type, break out 141 | case _ => fda_break 142 | } 143 | } 144 | ``` 145 | 146 | ### defining user-defined-task 147 | 148 | As a functional stream, It seems that some of the data access and processing in *FunDA* could be achieved in some pure functional ways like the following: 149 | 150 | ``` 151 | fdaStream.map(row => transformData(row)).map(action => runQueryAction(action)) 152 | ``` 153 | unfortunately the fact is pure stream combinators lack powerful and flexible flow control abilities that are so crucial for processing stream elements, therefore **user-defined-task** is introduced as a programming model to deal with the situation. 154 | User-defined-tasks are functional combinators designed by users each to achieve a single minimal task, and a much more complexed final task could be assembled by composing many of these tiny tasks in a specific order and then ***startRun***. The signature of *FDAUserTask[ROW]* is as follows: 155 | 156 | ``` 157 | type FDAUserTask[ROW] = (ROW) => (Option[List[ROW]]) 158 | 159 | ``` 160 | the above reads: an **user-defined-task** takes a row as input, use or transform it, and as a way of flow control, signify the state of next step of stream by returning **Option[List[ROW]]** as a result of execution of fda_next, fda_skip or fda_break. With the strong-typed-row requirement in place, the involved row types must extend from **FDAROW** and could be either a data-row or action-row. 161 | ##### types of rows 162 | *FunDA* streams are strong-typed, all rows must extend **FDAROW**. There are several categories of rows: 163 | 164 | * data-row: any case class extending **FDAROW** with parameters representing fields: 165 | **case class TypedRow(year: Int, state: String, value: Int) extends FDAROW** 166 | * action-row: case class extending **FDAROW** with a **Slick DBIOAction** wrapped inside the parameter as follows: 167 | **case class FDAActionRow(action: FDAAction) extends FDAROW** 168 | sometimes we need to target an action row to be run in different database context. In that case we can just define any case class and extend **FDAROW**: 169 | **case class MyActionRow(action: FDAAction) extends FDAROW** 170 | * error-row: case class extending **FDAROW** with a caught Exception object wrapped inside its parameter. 171 | **case class FDAErrorRow(e: Exception) extends FDAROW** 172 | user can define their own error row for different exceptions as long as they extend **FDAROW**: 173 | **case class MyErrorRow(msg: String, e: Exception) extends FDAROW** 174 | * null-row: a signal object used to represent EOS(end-of-stream): 175 | **case object FDANullRow extends FDAROW** 176 | 177 | ##### standard-operation-procedures 178 | User-defined-tasks have standard operation procedures as the following: 179 | 180 | 1. determine row type by pattern-matching 181 | 2. use row fields to perform data processing and transformation 182 | 3. control flow of rows downstream 183 | 184 | the following are samples of a few different purposed user-defined-tasks: 185 | 186 | ``` 187 | //strong typed row type. must extend FDAROW 188 | case class StateRow(state: String) extends FDAROW 189 | 190 | //a logging task. show name and pass row untouched downstream 191 | def showState: FDAUserTask[FDAROW] = row => { 192 | row match { 193 | case StateRow(sname) => //this is my row 194 | println(s"Name of state is：$sname") 195 | fda_next(row) 196 | case _ => fda_skip //not my row, do not pass it 197 | } 198 | } 199 | ``` 200 | 201 | ``` 202 | //a filter and processing task. 203 | //filter out rows with inconvertible value strings and out of ranged value and year 204 | def filterRows: FDAUserTask[FDAROW] = row => { 205 | row match { 206 | case r: AQMRaw => { //this is the correct row 207 | try { //process this row and catch exceptions 208 | val yr = r.year.toInt 209 | val v = r.value.toInt 210 | val vlu = if ( v > 10 ) 10 else v //max value allowed 211 | //construct a new row 212 | val data = AQMRPTModel(0,r.mid.toInt,r.state,r.county,yr,vlu,0,true) 213 | if ((yr > 1960 && yr < 2018)) //filtering 214 | fda_next(data) //this row ok. pass downstream 215 | else 216 | fda_skip //filter out this row 217 | } catch { 218 | case e: Exception => 219 | fda_next(FDAErrorRow(e)) //pass the caught exception as a row downstream 220 | } 221 | } 222 | case _ => fda_skip //wrong type, skip 223 | } 224 | } 225 | ``` 226 | 227 | ``` 228 | //a row transformation task 229 | //transform data to action for later execution 230 | def toAction: FDAUserTask[FDAROW] = row => { 231 | row match { 232 | case r: AQMRPTModel => //this is my row 233 | val queryAction = AQMRPTQuery += r //slick action 234 | fda_next(FDAActionRow(queryAction)) 235 | case _ => fda_skip 236 | } 237 | } 238 | ``` 239 | 240 | ``` 241 | //a query action runner task 242 | //get a query runner and an action task 243 | val actionRunner = FDAActionRunner(slick.jdbc.H2Profile) 244 | def runActionRow: FDAUserTask[FDAROW] = action => { 245 | action match { 246 | case FDAActionRow(q) => //this is a query action row 247 | actionRunner.fda_execAction(q)(db) //run it 248 | fda_skip 249 | case other@_ => fda_next(other) //don't touch it, just pass down 250 | //someone else downstream could process it 251 | } 252 | } 253 | 254 | ``` 255 | 256 | to run many task as a whole, we compose them and **startRun**: 257 | 258 | ``` 259 | //compose the program 260 | val streamAllTasks = streamAQMRaw.appendTask(filterRows) 261 | .appendTask(toAction) 262 | .appendTask(runActionRow) 263 | //run program 264 | streamToRun.startRun 265 | 266 | ``` 267 | ##### aggregation 268 | In stream style processing, many times we need to aggregate over rows, this is where **user-aggregate-task** is designed to fit in. An **user-aggregate-task** has the following signature: 269 | ``` 270 | type FDAAggrTask[AGGR,ROW] = (AGGR,ROW) => (AGGR,Option[List[ROW]]) 271 | ``` 272 | *AGGR* could be any user defined type to represent the state of aggregation. From the above type signature, we can see it is a typical functional style function represented by input a state and output new state. The following is an example of **user-aggregate-task**: 273 | 274 | ``` 275 | //define a structure to represent aggregator type 276 | case class Accu(state: String, county: String, year: Int, count: Int, sumOfValue: Int) 277 | 278 | //user defined aggregation task. only pass aggregated row downstream 279 | def countingAverage: FDAAggrTask[Accu,FDAROW] = (accu,row) => { 280 | row match { 281 | case aqmr: AQMRPTModel => //this is the target row type 282 | if (accu.state == "" || (aqmr.state == accu.state && aqmr.year == accu.year)) 283 | //same condition: inc count and add sum, no need to pass row downstream 284 | (Accu(aqmr.state,aqmr.county,aqmr.year,accu.count+1, accu.sumOfValue+aqmr.value),fda_skip) 285 | else 286 | //reset accumulator, create a new aggregated row and pass downstream 287 | (Accu(aqmr.state,aqmr.county,aqmr.year,1, aqmr.value) 288 | ,fda_next(AQMRPTModel(0,9999,accu.state,accu.county,accu.year 289 | ,accu.count,accu.sumOfValue/accu.count,true))) 290 | case FDANullRow => 291 | //last row encountered. create and pass new aggregated row 292 | (Accu(accu.state,accu.county,accu.year,1, 0) 293 | ,fda_next(AQMRPTModel(0,9999,accu.state,accu.county,accu.year 294 | ,accu.count,accu.sumOfValue/accu.count,true))) 295 | //incorrect row type, do nothing 296 | case _ => (accu,fda_skip) 297 | } 298 | } 299 | 300 | ``` 301 | 302 | the following demonstrates how it is executed: 303 | 304 | ``` 305 | aqmrStream.aggregateTask(Accu("","",0,0,0),countingAverage) 306 | .appendTask(toAction) 307 | .appendTask(runActionRow) 308 | .startRun 309 | 310 | ``` 311 | "aqmrStream" is a **source** with rows to be aggregated. 312 | ### Running programs inside user-defined-task 313 | A *FunDA* program consist of a **source** and multiple **user-defined-tasks**. It is possible to execute a *FunDA* program inside these *user_defined-tasks*. This means we have to call **startRun** inside the *user-defined-task* and some effect would inevitably be produced rending the calling *user-defined-task* impure. A complete example of *FunDA* program inside a *user-defined-task* is given below: 314 | 315 | ``` 316 | //getting id with corresponding name from STATES table 317 | def getStateID(state: String): Int = { 318 | //create a stream for state id with state name 319 | implicit def toState(row: StateTable#TableElementType) = StateModel(row.id,row.name) 320 | val stateLoader = FDAViewLoader(slick.jdbc.H2Profile)(toState _) 321 | val stateSeq = stateLoader.fda_typedRows(StateQuery.result)(db).toSeq 322 | //constructed a Stream[Task,String] 323 | val stateStream = fda_staticSource(stateSeq)() 324 | var id = -1 325 | def getid: FDAUserTask[FDAROW] = row => { 326 | row match { 327 | case StateModel(stid,stname) => //target row type 328 | if (stname.contains(state)) { 329 | id = stid 330 | fda_break //exit 331 | } 332 | else fda_skip //take next row 333 | case _ => fda_skip 334 | } 335 | } 336 | stateStream.appendTask(getid).startRun 337 | id 338 | } 339 | 340 | ``` 341 | "getStateID" is a *user-defined-function* in which function "getid" is physically executed. Because "getid" is local, we are still confident to use the calling *user-defined-function* in composition with other combinators as following: 342 | 343 | ``` 344 | //process input row and produce action row to insert into NORMAQM 345 | def getIdsThenInsertAction: FDAUserTask[FDAROW] = row => { 346 | row match { 347 | case aqm: AQMRPTModel => 348 | if (aqm.valid) { 349 | val stateId = getStateID(aqm.state) 350 | val countyId = getCountyID(aqm.state,aqm.county) 351 | val action = NORMAQMQuery += NORMAQMModel(0,aqm.mid, stateId, countyId, aqm.year,aqm.value,aqm.total) 352 | fda_next(FDAActionRow(action)) 353 | } 354 | else fda_skip 355 | case _ => fda_skip 356 | } 357 | } 358 | ``` 359 | in this case "getStateID" is called within another *user-defined-task*. 360 | ### Parallel Processing 361 | *FunDA* borrows its parallelism capabilities from *scalaz-streams-fs2*. There are two areas of parallel data processing application: 362 | 363 | 364 | * parallel loading of multiple **sources** 365 | * parallel execution of a single **user-defined-task** 366 | 367 | ##### parallel loading 368 | Parallel loading of many sources is achieved by calling function **fda-par-load** provided in *FunDA*. These sources could be constructed from tables on many separate database-servers or by spliting huge data tables into smaller un-repeated data chunks like the following: 369 | 370 | ``` 371 | //define query for extracting State names from AQMRPT 372 | val qryStates = AQMRPTQuery.map(_.state).distinct.sorted 373 | case class States(name: String) extends FDAROW 374 | implicit def toStates(row: String) = States(row) 375 | val stateLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toStates _) 376 | val statesStream = stateLoader.fda_typedStream(qryStates.result)(db_a)(64,64)() 377 | 378 | 379 | //define query for extracting County names from AQMRPT in separate chunks 380 | //query with state name >A and (r.state.toUpperCase > "A" && 382 | r.state.toUpperCase < "K")).map(r => (r.state,r.county)) 383 | .distinctOn(r => (r._1,r._2)) 384 | .sortBy(r => (r._1,r._2)) 385 | 386 | //query with state name >K and

(r.state.toUpperCase > "K" && 388 | r.state.toUpperCase < "P")).map(r => (r.state,r.county)) 389 | .distinctOn(r => (r._1,r._2)) 390 | .sortBy(r => (r._1,r._2)) 391 | 392 | //query with state name >P 393 | val qryCountiesP_Z = AQMRPTQuery.filter(r => r.state.toUpperCase > "P") 394 | .map(r => (r.state,r.county)) 395 | .distinctOn(r => (r._1,r._2)) 396 | .sortBy(r => (r._1,r._2)) 397 | 398 | case class Counties(state: String, name: String) extends FDAROW 399 | implicit def toCounties(row: (String,String)) = Counties(row._1,row._2) 400 | val countyLoader = FDAStreamLoader(slick.jdbc.H2Profile)(toCounties _) 401 | //3 separate streams to extract county names from the same database table AQMRPT 402 | val countiesA_KStream: FDAPipeLine[County] = countyLoader.fda_typedStream(qryCountiesA_K.result)(db_b)(64,64)() 403 | val countiesK_PStream: FDAPipeLine[County] = countyLoader.fda_typedStream(qryCountiesK_P.result)(db_b)(64,64)() 404 | val countiesP_ZStream: FDAPipeLine[County] = countyLoader.fda_typedStream(qryCountiesP_Z.result)(db_b)(64,64)() 405 | 406 | ``` 407 | once these **sources** are all constructed, we then load them parallelly: 408 | 409 | ``` 410 | //obtain a combined stream with parallel loading with max of 4 open computation 411 | val combinedStream: FDAPipeLine[FDAROW] = fda_par_load(statesStream,countiesA_KStream,countiesK_PStream,countiesP_ZStream)(4) 412 | 413 | ``` 414 | doing parallel loading would most likely to produce a stream with multiple types of rows, in the above case **States** and **Counties** represent two different types of rows respectively. Therefore *user-defined-tasks* each targeting different type of row are designed to handle rows of its target type, like the following: 415 | 416 | ``` 417 | //user-task to catch rows of States type and transform them into db insert actions 418 | def processStates: FDAUserTask[FDAROW] = row => { 419 | row match { 420 | //catch states row and transform it into insert action 421 | case States(stateName) => //target row type 422 | println(s"State name: ${stateName}") 423 | val action = StateQuery += StateModel(0,stateName) 424 | fda_next(StateActionRow(action)) 425 | case others@ _ => //pass other types to next user-defined-tasks 426 | fda_next(others) 427 | } 428 | } 429 | //user-task to catch rows of Counties type and transform them into db insert actions 430 | def processCounties: FDAUserTask[FDAROW] = row => { 431 | row match { 432 | //catch counties row and transform it into insert action 433 | case Counties(stateName,countyName) => //target row type 434 | println(s"County ${countyName} of ${stateName}") 435 | val action = CountyQuery += CountyModel(0,countyName+ " of "+stateName) 436 | fda_next(CountyActionRow(action)) 437 | case others@ _ => //pass other types to next user-defined-tasks 438 | fda_next(others) 439 | } 440 | } 441 | ``` 442 | ###### parallel loading a stream of sources 443 | The above demonstration of parallel loading started with known number of sources. This is especially convenient for users to manually arrange sources of different row types in a parallel loading operation. But, when a list of sources is itself a stream, then to parallelly load the stream of sources we need to first convert the stream into **FDAParSource** over a **FDASourceLoader** function as follows: 444 | 445 | ``` 446 | //loading rows with year yr 447 | def loadRowsInYear(yr: Int) = { 448 | //a new query 449 | val query = AQMRPTQuery.filter(row => row.year === yr) 450 | //reuse same loader 451 | AQMRPTLoader.fda_typedStream(query.result)(db)(256, 256)(println(s"End of stream ${yr}!!!!!!")) 452 | } 453 | 454 | //loading rows by year 455 | def loadRowsByYear: FDASourceLoader = row => { 456 | row match { 457 | case Years(y) => loadRowsInYear(y) //produce stream of the year 458 | case _ => fda_appendRow(FDANullRow) 459 | } 460 | } 461 | 462 | //get parallel source constructor 463 | val parSource: FDAParSource = yearStream.toParSource(loadRowsByYear) 464 | 465 | ``` 466 | the following demonstrates loading of this parallel source: 467 | 468 | ``` 469 | //produce a stream from parallel source 470 | val stream: FDAPipeLine[FDAROW] = fda_par_source(parSource)(4) 471 | ``` 472 | **fda_par_source** is actually a parallel execution function analogous to **fda_runPar** which is described in the following section. 473 | ##### parallel execution 474 | *FunDA* provides a function **fda_runPar** as a parallel task runner. A parallel task has the following signature: 475 | 476 | ``` 477 | /** Parallel task type 478 | * stream of streams type for parallel running user action 479 | * use stream.toPar to convert from FDAUserTask 480 | */ 481 | type FDAParTask = Stream[Task,Stream[Task,Option[List[FDAROW]]]] 482 | 483 | ``` 484 | and a **FDAUserTask** can be converted to **FDAParTask** as bellow: 485 | 486 | ``` 487 | AQMRPTStream.toPar(getIdsThenInsertAction) 488 | ``` 489 | where "AQMRPTStream" is a *FunDA* **source** and **toPar** is its method to turn "getIdsThenInsertAction" into a parallel task of many instances running in different threads. The principle of parallel execution is by scrambling rows in a single input stream into several un-ordered streams as inputs to many instances of a single task running parallelly in different threads. A **FDAParTask** requires a special runner to be carried out as shown below: 490 | 491 | ``` 492 | fda_runPar(AQMRPTStream.toPar(getIdsThenInsertAction))(8) 493 | ``` 494 | **fda_runPar** has a signature as follows: 495 | 496 | ``` 497 | def fda_runPar(parTask: FDAParTask)(maxOpen: Int): FDAPipleLine[FDAROW] 498 | 499 | ``` 500 | **maxOpen** designates the maximum number of open computations or degree of parallelism and the actual number of open computations depends on a number of factors including CPU cores, size of thread-pool and no more than user suggested maximum number of open computations. Thread-pool can be adjusted from default values by declaring implicit instance of Strategy: 501 | 502 | ``` 503 | implicit val strategy = Strategy.fromCachedDaemonPool("cachedPool") 504 | // implicit val strategy = Strategy.fromFixedDaemonPool(6) 505 | 506 | ``` 507 | the actual performance of parallelism requires thorough tuning of thread-pool strategies with respect to number of CPU cores. For whatever configurations, the performance gain through parallelism over single-thread task demonstrates great significance. A complete composition of parallel loading and parallel execution has the following model: 508 | 509 | ``` 510 | //get parallel source constructor 511 | val parSource = yearStream.toParSource(loadRowsByYear) 512 | //implicit val strategy = Strategy.fromCachedDaemonPool("cachedPool") 513 | //produce a stream from parallel sources 514 | val source = fda_par_source(parSource)(4) 515 | //turn getIdsThenInsertAction into parallel task 516 | val parTasks = source.toPar(getIdsThenInsertAction) 517 | //runPar to produce a new stream 518 | val actionStream =fda_runPar(parTasks)(4) 519 | //turn runInsertAction into parallel task 520 | val parRun = actionStream.toPar(runInsertAction) 521 | //runPar and carry out by startRun 522 | fda_runPar(parRun)(2).startRun 523 | 524 | ``` 525 | ##### remarks when writing parallel processing programs 526 | According to experiments, **FunDA** concurrent combinators are quite sensitive to thread management. The **maxOpen** parameters of both **fda_runPar** and **fda_par_source** in the examples given above had to be tuned to avoid thread contention and process-hang. The example used **HikariCP** with parameters set below: 527 | 528 | ``` 529 | h2db { 530 | url = "jdbc:h2:tcp://localhost/~/slickdemo;mv_store=false" 531 | driver = "org.h2.Driver" 532 | connectionPool = HikariCP 533 | numThreads = 48 534 | maxConnections = 48 535 | minConnections = 12 536 | keepAliveConnection = true 537 | } 538 | 539 | ``` 540 | ### Exceptions handling and Finalizers 541 | **FunDA** provides a mechanism that guarantees a **finalizer** be called upon termination of the stream, no matter if it is naturally end of stream or break-out caused by interruptions or exceptions. **Finalizers** are in fact call-back-functions hooked-up to a *FunDA* program during source construction, like the following: 542 | 543 | ``` 544 | val view = fda_staticSource(stateSeq)(println("***Finally*** the end of view!!!")) 545 | val stream = streamLoader.fda_typedStream(StateQuery.result)(db)(64,64)(println("***Finally*** the end of stream!!!")) 546 | 547 | ``` 548 | exceptions can be caught by **onError** call-backs that are hooked-up at the **very end** of **FunDA** stream in order to catch exceptions from all work-nodes as follows: 549 | 550 | ``` 551 | val v = viewState.appendTask(errorRow).appendTask(trackRows) 552 | val v1 = v.onError {case e: Exception => println(s"Caught Error in viewState!!![${e.getMessage}]"); fda_appendRow(FDANullRow)} 553 | v1.startRun 554 | 555 | val s = streamState.appendTask(errorRow).appendTask(trackRows) 556 | val s1 = s.onError {case e: Exception => println(s"Caught Error in streamState!!![${e.getMessage}]"); fda_appendRow(FDANullRow)} 557 | s1.startRun 558 | 559 | ``` 560 | ##### user defined exceptions 561 | Sometimes we wish to watch some particular events and take corresponding actions when they take place. This can be achieved by user-defined-exceptions. User-defined-exceptions are special rows extending from **FDAROW** that can be caught by pattern matching. The following is an example of user-defined-exception and its handling: 562 | 563 | ``` 564 | case class DivideZeroError(msg: String, e: Exception) extends FDAROW 565 | def catchError: FDAUserTask[FDAROW] = row => { 566 | row match { 567 | case StateModel(id,name) => 568 | try { 569 | val idx = id / (id - 3) 570 | fda_next(StateModel(idx, name)) 571 | } catch { 572 | case e: Exception => //pass an error row 573 | fda_next(DivideZeroError(s"Divide by zero exception at ${id}",e)) 574 | } 575 | case m@_ => fda_next(m) 576 | } 577 | } 578 | 579 | def trackRows: FDAUserTask[FDAROW] = row => { 580 | row match { 581 | case m@StateModel(id,name) => 582 | println(s"State: $id $name") 583 | println( "----------------") 584 | fda_next(m) 585 | case DivideZeroError(msg, e) => //error row 586 | println(s"***Error:$msg***") 587 | fda_skip 588 | case m@_ => fda_next(m) 589 | } 590 | } 591 | 592 | val s = streamState.take(5).appendTask(catchError).appendTask(trackRows) 593 | val s1 = s.onError {case e: Exception => println(s"Caught Error in streamState!!![${e.getMessage}]"); fda_appendRow(FDANullRow)} 594 | s1.startRun 595 | 596 | ``` --------------------------------------------------------------------------------