Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance. Project price only 1 $
You can buy this project and download/modify it how often you want.
/*
*
* * Licensed to the Apache Software Foundation (ASF) under one or more
* * contributor license agreements. See the NOTICE file distributed with
* * this work for additional information regarding copyright ownership.
* * The ASF licenses this file to You under the Apache License, Version 2.0
* * (the "License"); you may not use this file except in compliance with
* * the License. You may obtain a copy of the License at
* *
* * http://www.apache.org/licenses/LICENSE-2.0
* *
* * Unless required by applicable law or agreed to in writing, software
* * distributed under the License is distributed on an "AS IS" BASIS,
* * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* * See the License for the specific language governing permissions and
* * limitations under the License.
*
*
*/
package com.ebiznext.comet.job.ingest
import java.sql.Timestamp
import java.time.Instant
import com.ebiznext.comet.config.Settings
import com.ebiznext.comet.schema.handlers.{SchemaHandler, StorageHandler}
import com.ebiznext.comet.schema.model.Rejection.{ColInfo, ColResult, RowInfo, RowResult}
import com.ebiznext.comet.schema.model._
import org.apache.hadoop.fs.Path
import org.apache.spark.rdd.RDD
import org.apache.spark.sql._
import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
import org.apache.spark.sql.types.StructType
import scala.reflect.runtime.universe
import scala.util.{Failure, Success, Try}
/**
* Main class to ingest delimiter separated values file
*
* @param domain : Input Dataset Domain
* @param schema : Input Dataset Schema
* @param types : List of globally defined types
* @param path : Input dataset path
* @param storageHandler : Storage Handler
*/
class DsvIngestionJob(
val domain: Domain,
val schema: Schema,
val types: List[Type],
val path: List[Path],
val storageHandler: StorageHandler,
val schemaHandler: SchemaHandler
)(implicit val settings: Settings)
extends IngestionJob {
/**
* @return Spark Job name
*/
override def name: String =
s"""${domain.name}-${schema.name}-${path.map(_.getName).mkString(",")}"""
/**
* dataset Header names as defined by the schema
*/
val schemaHeaders: List[String] = schema.attributes.map(_.name)
/**
* remove any extra quote / BOM in the header
*
* @param header : Header column name
* @return
*/
def cleanHeaderCol(header: String): String =
header.replaceAll("\"", "").replaceAll("\uFEFF", "")
/**
* @param datasetHeaders : Headers found in the dataset
* @param schemaHeaders : Headers defined in the schema
* @return success if all headers in the schema exist in the dataset
*/
def validateHeader(datasetHeaders: List[String], schemaHeaders: List[String]): Boolean = {
schemaHeaders.forall(schemaHeader => datasetHeaders.contains(schemaHeader))
}
/**
* @param datasetHeaders : Headers found in the dataset
* @param schemaHeaders : Headers defined in the schema
* @return two lists : One with thecolumns present in the schema and the dataset and onther with the headers present in the dataset only
*/
def intersectHeaders(
datasetHeaders: List[String],
schemaHeaders: List[String]
): (List[String], List[String]) = {
datasetHeaders.partition(schemaHeaders.contains)
}
/**
* Load dataset using spark csv reader and all metadata. Does not infer schema.
* columns not defined in the schema are dropped fro the dataset (require datsets with a header)
*
* @return Spark Dataset
*/
def loadDataSet(): Try[DataFrame] = {
try {
val df = session.read
.option("header", metadata.isWithHeader().toString)
.option("inferSchema", value = false)
.option("delimiter", metadata.getSeparator())
.option("multiLine", metadata.getMultiline())
.option("quote", metadata.getQuote())
.option("escape", metadata.getEscape())
.option("parserLib", "UNIVOCITY")
.option("encoding", metadata.getEncoding())
.csv(path.map(_.toString): _*)
logger.debug(df.schema.treeString)
val resDF = metadata.withHeader match {
case Some(true) =>
val datasetHeaders: List[String] = df.columns.toList.map(cleanHeaderCol)
val (_, drop) = intersectHeaders(datasetHeaders, schemaHeaders)
if (datasetHeaders.length == drop.length) {
throw new Exception(s"""No attribute found in input dataset ${path.toString}
| SchemaHeaders : ${schemaHeaders.mkString(",")}
| Dataset Headers : ${datasetHeaders.mkString(",")}
""".stripMargin)
}
df.drop(drop: _*)
case Some(false) | None =>
/* No header, let's make sure we take the first attributes
if there are more in the CSV file
*/
val attributesWithoutscript = schema.attributes.filter(!_.script.isDefined)
val compare =
attributesWithoutscript.length.compareTo(df.columns.length)
if (compare == 0) {
df.toDF(
attributesWithoutscript
.map(_.name)
.take(attributesWithoutscript.length): _*
)
} else if (compare > 0) {
val countMissing = attributesWithoutscript.length - df.columns.length
throw new Exception(s"$countMissing columns in the input DataFrame ")
} else { // compare < 0
val cols = df.columns
df.select(
cols.head,
cols.tail
.take(attributesWithoutscript.length - 1): _*
).toDF(attributesWithoutscript.map(_.name): _*)
}
}
Success(
resDF.withColumn(
// Spark here can detect the input file automatically, so we're just using the input_file_name spark function
Settings.cometInputFileNameColumn,
org.apache.spark.sql.functions.input_file_name()
)
)
} catch {
case e: Exception =>
Failure(e)
}
}
def rowValidator(): DsvValidator = {
val runtimeMirror = universe.runtimeMirror(getClass.getClassLoader)
val module = runtimeMirror.staticModule(settings.comet.rowValidatorClass)
val obj: universe.ModuleMirror = runtimeMirror.reflectModule(module)
obj.instance.asInstanceOf[DsvValidator]
}
/**
* Apply the schema to the dataset. This is where all the magic happen
* Valid records are stored in the accepted path / table and invalid records in the rejected path / table
*
* @param dataset : Spark Dataset
*/
def ingest(dataset: DataFrame): (RDD[_], RDD[_]) = {
val attributesWithoutscript: Seq[Attribute] =
schema.attributes.filter(!_.script.isDefined) :+ Attribute(
name = Settings.cometInputFileNameColumn
)
def reorderAttributes(): List[Attribute] = {
val attributesMap = attributesWithoutscript.map(attr => (attr.name, attr)).toMap
val cols = dataset.columns
cols.map(colName => attributesMap(colName)).toList
}
val orderedAttributes = reorderAttributes()
def reorderTypes(): (List[Type], StructType) = {
val typeMap: Map[String, Type] = types.map(tpe => tpe.name -> tpe).toMap
val (tpes, sparkFields) = orderedAttributes.map { attribute =>
val tpe = typeMap(attribute.`type`)
(tpe, tpe.sparkType(attribute.name, !attribute.required, attribute.comment))
}.unzip
(tpes, StructType(sparkFields))
}
val (orderedTypes, orderedSparkTypes) = reorderTypes()
val (rejectedRDD, acceptedRDD) = rowValidator().validate(
session,
dataset,
orderedAttributes,
orderedTypes,
orderedSparkTypes
)
saveRejected(rejectedRDD)
val (df, _) = saveAccepted(acceptedRDD, orderedSparkTypes)
index(df)
(rejectedRDD, acceptedRDD)
}
def saveAccepted(acceptedRDD: RDD[Row], orderedSparkTypes: StructType): (DataFrame, Path) = {
val renamedAttributes = schema.renamedAttributes().toMap
logger.whenInfoEnabled {
renamedAttributes.foreach {
case (name, rename) =>
logger.info(s"renaming column $name to $rename")
}
}
val acceptedDF = session.createDataFrame(acceptedRDD, orderedSparkTypes)
val finalDF =
renamedAttributes.foldLeft(acceptedDF)((acc, ca) => acc.withColumnRenamed(ca._1, ca._2))
super.saveAccepted(finalDF)
}
}
trait DsvValidator {
/**
* For each col of each row
* - we extract the col value / the col constraints / col type
* - we check that the constraints are verified
* - we apply any required privacy transformation
* - parse the column into the target primitive Spark Type
* We end up using catalyst to create a Spark Row
*
* @param session : The Spark session
* @param dataset : The dataset
* @param attributes : the col attributes
* @param types : List of globally defined types
* @param sparkType : The expected Spark Type for valid rows
* @return Two RDDs : One RDD for rejected rows and one RDD for accepted rows
*/
def validate(
session: SparkSession,
dataset: DataFrame,
attributes: List[Attribute],
types: List[Type],
sparkType: StructType
)(implicit settings: Settings): (RDD[String], RDD[Row])
}
/**
* The Spark task that run on each worker
*/
object DsvIngestionUtil extends DsvValidator {
override def validate(
session: SparkSession,
dataset: DataFrame,
attributes: List[Attribute],
types: List[Type],
sparkType: StructType
)(implicit settings: Settings): (RDD[String], RDD[Row]) = {
val now = Timestamp.from(Instant.now)
val checkedRDD: RDD[RowResult] = dataset.rdd
.mapPartitions { partition =>
partition.map { row: Row =>
val rowValues: Seq[(String, Attribute)] = row.toSeq
.zip(attributes)
.map {
case (colValue, colAttribute) =>
(Option(colValue).getOrElse("").toString, colAttribute)
}
val rowCols = rowValues.zip(types)
val colMap = rowValues.map(__ => (__._2.name, __._1)).toMap
val validNumberOfColumns = attributes.length <= rowCols.length
if (!validNumberOfColumns) {
RowResult(
rowCols.map {
case ((colRawValue, colAttribute), tpe) =>
ColResult(
ColInfo(
colRawValue,
colAttribute.name,
tpe.name,
tpe.pattern,
false
),
null
)
}.toList
)
} else {
RowResult(
rowCols.map {
case ((colRawValue, colAttribute), tpe) =>
IngestionUtil.validateCol(colRawValue, colAttribute, tpe, colMap)
}.toList
)
}
}
} persist (settings.comet.cacheStorageLevel)
val rejectedRDD: RDD[String] = checkedRDD
.filter(_.isRejected)
.map(rr => RowInfo(now, rr.colResults.filter(!_.colInfo.success).map(_.colInfo)).toString)
val acceptedRDD: RDD[Row] = checkedRDD.filter(_.isAccepted).map { rowResult =>
val sparkValues: List[Any] = rowResult.colResults.map(_.sparkValue)
new GenericRowWithSchema(Row(sparkValues: _*).toSeq.toArray, sparkType)
}
(rejectedRDD, acceptedRDD)
}
}
object AcceptAllValidator extends DsvValidator {
override def validate(
session: SparkSession,
dataset: DataFrame,
attributes: List[Attribute],
types: List[Type],
sparkType: StructType
)(implicit settings: Settings): (RDD[String], RDD[Row]) = {
val rejectedRDD: RDD[String] = session.emptyDataFrame.rdd.map(_.mkString)
val acceptedRDD: RDD[Row] = dataset.rdd
(rejectedRDD, acceptedRDD)
}
}
