forked from apache/carbondata
-
Notifications
You must be signed in to change notification settings - Fork 0
/
CarbonScanPartitionRDD.scala
266 lines (251 loc) · 11.4 KB
/
CarbonScanPartitionRDD.scala
1
2
3
4
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
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
/*
* 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 org.apache.carbondata.spark.rdd
import java.util.ArrayList
import scala.collection.JavaConverters._
import scala.collection.mutable.ArrayBuffer
import org.apache.hadoop.conf.Configuration
import org.apache.hadoop.mapred.JobConf
import org.apache.hadoop.mapreduce.Job
import org.apache.spark.{Partition, TaskContext}
import org.apache.spark.rdd.RDD
import org.apache.spark.sql.execution.command.AlterPartitionModel
import org.apache.spark.sql.hive.DistributionUtil
import org.apache.spark.sql.util.CarbonException
import org.apache.spark.unsafe.types.UTF8String
import org.apache.spark.util.PartitionUtils
import org.apache.carbondata.common.logging.LogServiceFactory
import org.apache.carbondata.core.datastore.block.{Distributable, SegmentProperties, TaskBlockInfo}
import org.apache.carbondata.core.keygenerator.directdictionary.DirectDictionaryKeyGeneratorFactory
import org.apache.carbondata.core.metadata.AbsoluteTableIdentifier
import org.apache.carbondata.core.metadata.datatype.DataTypes
import org.apache.carbondata.core.metadata.encoder.Encoding
import org.apache.carbondata.core.metadata.schema.partition.PartitionType
import org.apache.carbondata.core.metadata.schema.table.column.{CarbonDimension, CarbonMeasure}
import org.apache.carbondata.core.scan.result.iterator.PartitionSpliterRawResultIterator
import org.apache.carbondata.core.scan.wrappers.ByteArrayWrapper
import org.apache.carbondata.core.util.{ByteUtil, DataTypeUtil}
import org.apache.carbondata.hadoop.{CarbonInputSplit, CarbonMultiBlockSplit}
import org.apache.carbondata.hadoop.util.CarbonInputFormatUtil
import org.apache.carbondata.processing.merger.CarbonCompactionUtil
import org.apache.carbondata.processing.partition.spliter.CarbonSplitExecutor
import org.apache.carbondata.processing.util.CarbonLoaderUtil
/**
* This RDD is used in alter table partition statement to get data of target partitions,
* then repartition data according to new partitionInfo
* @param alterPartitionModel
* @param absoluteTableIdentifier
* @param partitionIds the ids of target partition to be scanned
* @param bucketId
*/
class CarbonScanPartitionRDD(alterPartitionModel: AlterPartitionModel,
absoluteTableIdentifier: AbsoluteTableIdentifier,
partitionIds: Seq[String],
bucketId: Int)
extends RDD[(AnyRef, Array[AnyRef])](alterPartitionModel.sqlContext.sparkContext, Nil) {
private val queryId = alterPartitionModel.sqlContext.sparkContext.getConf
.get("queryId", System.nanoTime() + "")
val segmentId = alterPartitionModel.segmentId
val carbonLoadModel = alterPartitionModel.carbonLoadModel
val oldPartitionIdList = alterPartitionModel.oldPartitionIds
var storeLocation: String = null
var splitStatus: Boolean = false
var blockId: String = null
val carbonTable = carbonLoadModel.getCarbonDataLoadSchema.getCarbonTable
val dimensions = carbonTable.getAllDimensions.asScala
val measures = carbonTable.getAllMeasures.asScala
val partitionInfo = carbonTable
.getPartitionInfo(absoluteTableIdentifier.getCarbonTableIdentifier.getTableName)
val partitionColumn = partitionInfo.getColumnSchemaList().get(0)
val partitionDataType = partitionColumn.getDataType
val partitionColumnName = partitionColumn.getColumnName
var isDimension: Boolean = false
val encodingList = partitionColumn.getEncodingList
var dimension: CarbonDimension = null
var measure: CarbonMeasure = null
val noDictionaryIndexGroup: ArrayBuffer[Int] = new ArrayBuffer[Int]()
val dictionaryIndexGroup: ArrayBuffer[Int] = new ArrayBuffer[Int]()
val measureIndexGroup: ArrayBuffer[Int] = new ArrayBuffer[Int]()
override def getPartitions: Array[Partition] = {
val parallelism = sparkContext.defaultParallelism
val jobConf = new JobConf(new Configuration)
val job = new Job(jobConf)
val format = CarbonInputFormatUtil.createCarbonTableInputFormat(absoluteTableIdentifier,
partitionIds.toList.asJava, job)
job.getConfiguration.set("query.id", queryId)
val splits = format.getSplitsOfOneSegment(job, segmentId,
oldPartitionIdList.map(_.asInstanceOf[Integer]).asJava, partitionInfo)
var partition_num = 0
val result = new ArrayList[Partition](parallelism)
val blockList = splits.asScala
.filter(_.asInstanceOf[CarbonInputSplit].getBucketId.toInt == bucketId)
.map(_.asInstanceOf[Distributable])
if (!blockList.isEmpty) {
val activeNodes = DistributionUtil.ensureExecutorsAndGetNodeList(blockList, sparkContext)
val nodeBlockMapping = CarbonLoaderUtil.nodeBlockTaskMapping(blockList.asJava, -1,
parallelism, activeNodes.toList.asJava)
nodeBlockMapping.asScala.foreach { case (node, blockList) =>
blockList.asScala.foreach { blocksPerTask =>
val splits = blocksPerTask.asScala.map(_.asInstanceOf[CarbonInputSplit])
if (blocksPerTask.size() != 0) {
val multiBlockSplit =
new CarbonMultiBlockSplit(absoluteTableIdentifier, splits.asJava, Array(node))
val partition = new CarbonSparkPartition(id, partition_num, multiBlockSplit)
result.add(partition)
partition_num += 1
}
}
}
}
result.toArray(new Array[Partition](result.size()))
}
override def compute(split: Partition, context: TaskContext):
Iterator[(AnyRef, Array[AnyRef])] = {
val LOGGER = LogServiceFactory.getLogService(this.getClass.getName)
var exec : CarbonSplitExecutor = null
val rows : java.util.List[(AnyRef, Array[AnyRef])] = new ArrayList[(AnyRef, Array[AnyRef])]()
try {
val inputSplit = split.asInstanceOf[CarbonSparkPartition].split.value
val splits = inputSplit.getAllSplits.asScala
val tableBlockInfoList = CarbonInputSplit.createBlocks(splits.asJava)
val segmentMapping: java.util.Map[String, TaskBlockInfo] =
CarbonCompactionUtil.createMappingForSegments(tableBlockInfoList)
val carbonTable = carbonLoadModel.getCarbonDataLoadSchema.getCarbonTable
var result : java.util.List[PartitionSpliterRawResultIterator] = null
try {
exec = new CarbonSplitExecutor(segmentMapping, carbonTable)
result = exec.processDataBlocks(segmentId)
} catch {
case e: Throwable =>
LOGGER.error(e)
if (null != e.getMessage) {
CarbonException.analysisException(
s"Exception occurred in query execution :: ${e.getMessage}")
} else {
CarbonException.analysisException(
"Exception occurred in query execution. Please check logs.")
}
}
val segmentProperties = PartitionUtils.getSegmentProperties(absoluteTableIdentifier,
segmentId, partitionIds.toList, oldPartitionIdList, partitionInfo)
val partColIdx = getPartitionColumnIndex(partitionColumnName, segmentProperties)
indexInitialise()
for (iterator <- result.asScala) {
while (iterator.hasNext) {
val row = iterator.next()
val partitionColumnValue = getPartitionColumnValue(row, partColIdx,
segmentProperties)
rows.add((partitionColumnValue, row))
}
}
} catch {
case e: Exception =>
LOGGER.error(e)
throw e
} finally {
if (null != exec) {
exec.finish
}
}
val iter = rows.iterator().asScala
iter
}
def getPartitionColumnValue(row: Array[AnyRef], partColIdx: Int,
segmentProperties: SegmentProperties): AnyRef = {
val dims: Array[Byte] = row(0).asInstanceOf[ByteArrayWrapper].getDictionaryKey
val keyGen = segmentProperties.getDimensionKeyGenerator
val keyArray: Array[Long] = keyGen.getKeyArray(dims)
val encodings = partitionColumn.getEncodingList
val partitionType = partitionInfo.getPartitionType
var partitionValue: AnyRef = null
val factor = 1000L
if (isDimension) {
// direct dictionary
if (encodings.contains(Encoding.DIRECT_DICTIONARY)) {
val directDictionaryGenerator = DirectDictionaryKeyGeneratorFactory
.getDirectDictionaryGenerator(partitionDataType)
val dictionaryIndex = dictionaryIndexGroup.indexOf(partColIdx)
val surrogateValue = (keyArray(dictionaryIndex) / factor).toInt
partitionValue = directDictionaryGenerator.getValueFromSurrogate(surrogateValue)
} else if (!encodings.contains(Encoding.DICTIONARY)) {
// no dictionary
val byteArray = row(0).asInstanceOf[ByteArrayWrapper].getNoDictionaryKeys
val index = noDictionaryIndexGroup.indexOf(partColIdx)
partitionValue = DataTypeUtil.getDataBasedOnDataTypeForNoDictionaryColumn(byteArray(index)
, partitionDataType)
if (partitionValue.isInstanceOf[UTF8String]) {
partitionValue = partitionValue.toString
}
} else { // normal dictionary
val dict = CarbonLoaderUtil.getDictionary(absoluteTableIdentifier,
dimension.getColumnIdentifier, partitionDataType)
if (partitionDataType == DataTypes.STRING) {
if (partitionType == PartitionType.RANGE) {
partitionValue = ByteUtil.
toBytes(dict.getDictionaryValueForKey(keyArray(partColIdx).toInt))
} else {
partitionValue = dict.getDictionaryValueForKey(keyArray(partColIdx).toInt)
}
} else {
partitionValue = dict.getDictionaryValueForKey(keyArray(partColIdx).toInt)
}
}
} else {
partitionValue = row(measureIndexGroup(partColIdx))
}
partitionValue
}
def indexInitialise(): Unit = {
for (dim: CarbonDimension <- dimensions) {
if (!dim.getEncoder.contains(Encoding.DICTIONARY)) {
noDictionaryIndexGroup.append(dimensions.indexOf(dim))
} else {
dictionaryIndexGroup.append(dimensions.indexOf(dim))
}
}
for (msr: CarbonMeasure <- measures) {
// index of measure in row
measureIndexGroup.append(measures.indexOf(msr) + 1)
}
}
/**
* get the index of partition column in dimension/measure
* @param partitionColumnName
* @param segmentProperties
* @return
*/
def getPartitionColumnIndex(partitionColumnName: String,
segmentProperties: SegmentProperties): Int = {
val dimensions = segmentProperties.getDimensions
val measures = segmentProperties.getMeasures
val columns = dimensions.asScala.map(_.getColName) ++ measures.asScala.map(_.getColName)
var index = 0
for (i <- 0 until columns.size) {
if (columns(i) == partitionColumnName) {
index = i
}
}
if (index < dimensions.size()) {
isDimension = true
dimension = dimensions.get(index)
} else {
index = index - dimensions.size()
measure = measures.get(index)
}
index
}
}