apache / carbondata / 643

Build # 643

Build Type

push

jenkins

Committed by GitHub

Commit Message

<a href="https://github.com/apache/carbondata/commit/<a class=hub.com/apache/carbondata/commit/a2169132250407e885df4d662ba31b7abea72d8a">a21691322<a href="https://github.com/apache/carbondata/commit/a2169132250407e885df4d662ba31b7abea72d8a">">Merge </a><a class="double-link" href="https://github.com/apache/carbondata/commit/<a class="double-link" href="https://github.com/apache/carbondata/commit/552e6a3f39a44836cd311538cb28b89392040ef1">552e6a3f3</a>">552e6a3f3</a><a href="https://github.com/apache/carbondata/commit/a2169132250407e885df4d662ba31b7abea72d8a"> into e3eb03054">e3eb03054</a>

Run Details

29 of 29 new or added lines in 7 files covered. (100.0%)

59864 of 74293 relevant lines covered (80.58%)

1.07 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

75.63

/datamap/bloom/src/main/java/org/apache/carbondata/datamap/bloom/BloomCoarseGrainDataMapFactory.java

/*
 * 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.datamap.bloom;

import java.io.File;
import java.io.IOException;
import java.util.*;
import java.util.concurrent.ConcurrentHashMap;

import org.apache.carbondata.common.annotations.InterfaceAudience;
import org.apache.carbondata.common.exceptions.sql.MalformedDataMapCommandException;
import org.apache.carbondata.common.logging.LogService;
import org.apache.carbondata.common.logging.LogServiceFactory;
import org.apache.carbondata.core.cache.Cache;
import org.apache.carbondata.core.cache.CacheProvider;
import org.apache.carbondata.core.cache.CacheType;
import org.apache.carbondata.core.constants.CarbonV3DataFormatConstants;
import org.apache.carbondata.core.datamap.DataMapDistributable;
import org.apache.carbondata.core.datamap.DataMapLevel;
import org.apache.carbondata.core.datamap.DataMapMeta;
import org.apache.carbondata.core.datamap.Segment;
import org.apache.carbondata.core.datamap.dev.DataMapBuilder;
import org.apache.carbondata.core.datamap.dev.DataMapFactory;
import org.apache.carbondata.core.datamap.dev.DataMapWriter;
import org.apache.carbondata.core.datamap.dev.cgdatamap.CoarseGrainDataMap;
import org.apache.carbondata.core.datastore.block.SegmentProperties;
import org.apache.carbondata.core.datastore.filesystem.CarbonFile;
import org.apache.carbondata.core.datastore.impl.FileFactory;
import org.apache.carbondata.core.features.TableOperation;
import org.apache.carbondata.core.metadata.schema.table.CarbonTable;
import org.apache.carbondata.core.metadata.schema.table.DataMapSchema;
import org.apache.carbondata.core.metadata.schema.table.column.CarbonColumn;
import org.apache.carbondata.core.scan.filter.intf.ExpressionType;
import org.apache.carbondata.core.statusmanager.SegmentStatusManager;
import org.apache.carbondata.core.util.CarbonUtil;
import org.apache.carbondata.core.util.path.CarbonTablePath;
import org.apache.carbondata.events.Event;

import org.apache.commons.lang3.StringUtils;

/**
 * This class is for Bloom Filter for blocklet level
 */
@InterfaceAudience.Internal
public class BloomCoarseGrainDataMapFactory extends DataMapFactory<CoarseGrainDataMap> {
  private static final LogService LOGGER = LogServiceFactory.getLogService(
      BloomCoarseGrainDataMapFactory.class.getName());
  /**
   * property for size of bloom filter
   */
  private static final String BLOOM_SIZE = "bloom_size";
  /**
   * default size for bloom filter, cardinality of the column.
   */
  private static final int DEFAULT_BLOOM_FILTER_SIZE =
      CarbonV3DataFormatConstants.NUMBER_OF_ROWS_PER_BLOCKLET_COLUMN_PAGE_DEFAULT * 20;
  /**
   * property for fpp(false-positive-probability) of bloom filter
   */
  private static final String BLOOM_FPP = "bloom_fpp";
  /**
   * default value for fpp of bloom filter is 0.001%
   */
  private static final double DEFAULT_BLOOM_FILTER_FPP = 0.00001d;

  /**
   * property for compressing bloom while saving to disk.
   */
  private static final String COMPRESS_BLOOM = "bloom_compress";
  /**
   * Default value of compressing bloom while save to disk.
   */
  private static final boolean DEFAULT_BLOOM_COMPRESS = true;

  private DataMapMeta dataMapMeta;
  private String dataMapName;
  private int bloomFilterSize;
  private double bloomFilterFpp;
  private boolean bloomCompress;
  private Cache<BloomCacheKeyValue.CacheKey, BloomCacheKeyValue.CacheValue> cache;
  // segmentId -> list of index file
  private Map<String, Set<String>> segmentMap = new ConcurrentHashMap<>();

  public BloomCoarseGrainDataMapFactory(CarbonTable carbonTable, DataMapSchema dataMapSchema)
      throws MalformedDataMapCommandException {
    super(carbonTable, dataMapSchema);
    Objects.requireNonNull(carbonTable);
    Objects.requireNonNull(dataMapSchema);

    this.dataMapName = dataMapSchema.getDataMapName();

    List<CarbonColumn> indexedColumns = carbonTable.getIndexedColumns(dataMapSchema);
    this.bloomFilterSize = validateAndGetBloomFilterSize(dataMapSchema);
    this.bloomFilterFpp = validateAndGetBloomFilterFpp(dataMapSchema);
    this.bloomCompress = validateAndGetBloomCompress(dataMapSchema);
    List<ExpressionType> optimizedOperations = new ArrayList<ExpressionType>();
    // todo: support more optimize operations
    optimizedOperations.add(ExpressionType.EQUALS);
    optimizedOperations.add(ExpressionType.IN);
    this.dataMapMeta = new DataMapMeta(this.dataMapName, indexedColumns, optimizedOperations);
    LOGGER.info(String.format("DataMap %s works for %s with bloom size %d",
        this.dataMapName, this.dataMapMeta, this.bloomFilterSize));
    try {
      this.cache = CacheProvider.getInstance()
          .createCache(new CacheType("bloom_cache"), BloomDataMapCache.class.getName());
    } catch (Exception e) {
      LOGGER.error(e);
      throw new MalformedDataMapCommandException(e.getMessage());
    }
  }

  /**
   * validate Lucene DataMap BLOOM_SIZE
   * 1. BLOOM_SIZE property is optional, 32000 * 20 will be the default size.
   * 2. BLOOM_SIZE should be an integer that greater than 0
   */
  private int validateAndGetBloomFilterSize(DataMapSchema dmSchema)
      throws MalformedDataMapCommandException {
    String bloomFilterSizeStr = dmSchema.getProperties().get(BLOOM_SIZE);
    if (StringUtils.isBlank(bloomFilterSizeStr)) {
      LOGGER.warn(
          String.format("Bloom filter size is not configured for datamap %s, use default value %d",
              dataMapName, DEFAULT_BLOOM_FILTER_SIZE));
      return DEFAULT_BLOOM_FILTER_SIZE;
    }
    int bloomFilterSize;
    try {
      bloomFilterSize = Integer.parseInt(bloomFilterSizeStr);
    } catch (NumberFormatException e) {
      throw new MalformedDataMapCommandException(
          String.format("Invalid value of bloom filter size '%s', it should be an integer",
              bloomFilterSizeStr));
    }
    // todo: reconsider the boundaries of bloom filter size
    if (bloomFilterSize <= 0) {
      throw new MalformedDataMapCommandException(
          String.format("Invalid value of bloom filter size '%s', it should be greater than 0",
              bloomFilterSizeStr));
    }
    return bloomFilterSize;
  }

  /**
   * validate bloom DataMap BLOOM_FPP
   * 1. BLOOM_FPP property is optional, 0.00001 will be the default value.
   * 2. BLOOM_FPP should be (0, 1)
   */
  private double validateAndGetBloomFilterFpp(DataMapSchema dmSchema)
      throws MalformedDataMapCommandException {
    String bloomFilterFppStr = dmSchema.getProperties().get(BLOOM_FPP);
    if (StringUtils.isBlank(bloomFilterFppStr)) {
      LOGGER.warn(
          String.format("Bloom filter FPP is not configured for datamap %s, use default value %f",
              dataMapName, DEFAULT_BLOOM_FILTER_FPP));
      return DEFAULT_BLOOM_FILTER_FPP;
    }
    double bloomFilterFpp;
    try {
      bloomFilterFpp = Double.parseDouble(bloomFilterFppStr);
    } catch (NumberFormatException e) {
      throw new MalformedDataMapCommandException(
          String.format("Invalid value of bloom filter fpp '%s', it should be an numeric",
              bloomFilterFppStr));
    }
    if (bloomFilterFpp < 0 || bloomFilterFpp - 1 >= 0) {
      throw new MalformedDataMapCommandException(
          String.format("Invalid value of bloom filter fpp '%s', it should be in range 0~1",
              bloomFilterFppStr));
    }
    return bloomFilterFpp;
  }

  /**
   * validate bloom DataMap COMPRESS_BLOOM
   * Default value is true
   */
  private boolean validateAndGetBloomCompress(DataMapSchema dmSchema) {
    String bloomCompress = dmSchema.getProperties().get(COMPRESS_BLOOM);
    if (StringUtils.isBlank(bloomCompress)) {
      LOGGER.warn(
          String.format("Bloom compress is not configured for datamap %s, use default value %b",
              dataMapName, DEFAULT_BLOOM_COMPRESS));
      return DEFAULT_BLOOM_COMPRESS;
    }
    return Boolean.parseBoolean(bloomCompress);
  }

  @Override
  public DataMapWriter createWriter(Segment segment, String shardName,
      SegmentProperties segmentProperties) throws IOException {
    LOGGER.info(
        String.format("Data of BloomCoarseGranDataMap %s for table %s will be written to %s",
            this.dataMapName, getCarbonTable().getTableName() , shardName));
    return new BloomDataMapWriter(getCarbonTable().getTablePath(), this.dataMapName,
        this.dataMapMeta.getIndexedColumns(), segment, shardName, segmentProperties,
        this.bloomFilterSize, this.bloomFilterFpp, bloomCompress);
  }

  @Override
  public DataMapBuilder createBuilder(Segment segment, String shardName,
      SegmentProperties segmentProperties) throws IOException {
    return new BloomDataMapBuilder(getCarbonTable().getTablePath(), this.dataMapName,
        this.dataMapMeta.getIndexedColumns(), segment, shardName, segmentProperties,
        this.bloomFilterSize, this.bloomFilterFpp, bloomCompress);
  }

  /**
   * returns all shard directories of bloom index files for query
   * if bloom index files are merged we should get only one shard path
   */
  private Set<String> getAllShardPaths(String tablePath, String segmentId) {
    String dataMapStorePath = CarbonTablePath.getDataMapStorePath(
        tablePath, segmentId, dataMapName);
    CarbonFile[] carbonFiles = FileFactory.getCarbonFile(dataMapStorePath).listFiles();
    Set<String> shardPaths = new HashSet<>();
    boolean mergeShardInprogress = false;
    CarbonFile mergeShardFile = null;
    for (CarbonFile carbonFile : carbonFiles) {
      if (carbonFile.getName().equals(BloomIndexFileStore.MERGE_BLOOM_INDEX_SHARD_NAME)) {
        mergeShardFile = carbonFile;
      } else if (carbonFile.getName().equals(BloomIndexFileStore.MERGE_INPROGRESS_FILE)) {
        mergeShardInprogress = true;
      } else if (carbonFile.isDirectory()) {
        shardPaths.add(carbonFile.getAbsolutePath());
      }
    }
    if (mergeShardFile != null && !mergeShardInprogress) {
      // should only get one shard path if mergeShard is generated successfully
      shardPaths.clear();
      shardPaths.add(mergeShardFile.getAbsolutePath());
    }
    return shardPaths;
  }

  @Override
  public List<CoarseGrainDataMap> getDataMaps(Segment segment) throws IOException {
    List<CoarseGrainDataMap> dataMaps = new ArrayList<>();
    try {
      Set<String> shardPaths = segmentMap.get(segment.getSegmentNo());
      if (shardPaths == null) {
        shardPaths = getAllShardPaths(getCarbonTable().getTablePath(), segment.getSegmentNo());
        segmentMap.put(segment.getSegmentNo(), shardPaths);
      }
      Set<String> filteredShards = segment.getFilteredIndexShardNames();
      for (String shard : shardPaths) {
        if (shard.endsWith(BloomIndexFileStore.MERGE_BLOOM_INDEX_SHARD_NAME) ||
            filteredShards.contains(new File(shard).getName())) {
          // Filter out the tasks which are filtered through Main datamap.
          // for merge shard, shard pruning delay to be done before pruning blocklet
          BloomCoarseGrainDataMap bloomDM = new BloomCoarseGrainDataMap();
          bloomDM.init(new BloomDataMapModel(shard, cache, segment.getConfiguration()));
          String dataMapStorePath = CarbonTablePath.getDataMapStorePath(
              getCarbonTable().getTablePath(), segment.getSegmentNo(), dataMapName);
          bloomDM.initIndexColumnConverters(getCarbonTable(), dataMapStorePath,
              dataMapMeta.getIndexedColumns());
          bloomDM.setFilteredShard(filteredShards);
          dataMaps.add(bloomDM);
        }
      }
    } catch (Exception e) {
      throw new IOException("Error occurs while init Bloom DataMap", e);
    }
    return dataMaps;
  }

  @Override
  public List<CoarseGrainDataMap> getDataMaps(DataMapDistributable distributable)
      throws IOException {
    List<CoarseGrainDataMap> dataMaps = new ArrayList<>();
    String indexPath = ((BloomDataMapDistributable) distributable).getIndexPath();
    Set<String> filteredShards = ((BloomDataMapDistributable) distributable).getFilteredShards();
    BloomCoarseGrainDataMap bloomDM = new BloomCoarseGrainDataMap();
    bloomDM.init(new BloomDataMapModel(indexPath, cache, FileFactory.getConfiguration()));
    String dmStorePath = CarbonTablePath.getDataMapStorePath(
        getCarbonTable().getTablePath(), distributable.getSegment().getSegmentNo(), dataMapName);
    bloomDM.initIndexColumnConverters(getCarbonTable(), dmStorePath,
        dataMapMeta.getIndexedColumns());
    bloomDM.setFilteredShard(filteredShards);
    dataMaps.add(bloomDM);
    return dataMaps;
  }


  @Override
  public List<DataMapDistributable> toDistributable(Segment segment) {
    List<DataMapDistributable> dataMapDistributableList = new ArrayList<>();
    Set<String> shardPaths = segmentMap.get(segment.getSegmentNo());
    if (shardPaths == null) {
      shardPaths = getAllShardPaths(getCarbonTable().getTablePath(), segment.getSegmentNo());
      segmentMap.put(segment.getSegmentNo(), shardPaths);
    }
    Set<String> filteredShards = segment.getFilteredIndexShardNames();
    for (String shardPath : shardPaths) {
      // Filter out the tasks which are filtered through Main datamap.
      // for merge shard, shard pruning delay to be done before pruning blocklet
      if (shardPath.endsWith(BloomIndexFileStore.MERGE_BLOOM_INDEX_SHARD_NAME) ||
          filteredShards.contains(new File(shardPath).getName())) {
        DataMapDistributable bloomDataMapDistributable =
            new BloomDataMapDistributable(shardPath, filteredShards);
        dataMapDistributableList.add(bloomDataMapDistributable);
      }
    }
    return dataMapDistributableList;
  }

  @Override
  public void fireEvent(Event event) {

  }

  @Override
  public void clear(Segment segment) {
    Set<String> shards = segmentMap.remove(segment.getSegmentNo());
    if (shards != null) {
      for (String shard : shards) {
        for (CarbonColumn carbonColumn : dataMapMeta.getIndexedColumns()) {
          cache.invalidate(new BloomCacheKeyValue.CacheKey(shard, carbonColumn.getColName()));
        }
      }
    }
  }

  @Override
  public synchronized void clear() {
    if (segmentMap.size() > 0) {
      List<String> segments = new ArrayList<>(segmentMap.keySet());
      for (String segmentId : segments) {
        clear(new Segment(segmentId, null, null));
      }
    }
  }

  @Override
  public void deleteDatamapData(Segment segment) throws IOException {
    try {
      String segmentId = segment.getSegmentNo();
      String datamapPath = CarbonTablePath
          .getDataMapStorePath(getCarbonTable().getTablePath(), segmentId, dataMapName);
      if (FileFactory.isFileExist(datamapPath)) {
        CarbonFile file = FileFactory.getCarbonFile(datamapPath,
            FileFactory.getFileType(datamapPath));
        CarbonUtil.deleteFoldersAndFilesSilent(file);
      }
    } catch (InterruptedException ex) {
      throw new IOException("Failed to delete datamap for segment_" + segment.getSegmentNo());
    }
  }

  @Override
  public void deleteDatamapData() {
    SegmentStatusManager ssm =
        new SegmentStatusManager(getCarbonTable().getAbsoluteTableIdentifier());
    try {
      List<Segment> validSegments = ssm.getValidAndInvalidSegments().getValidSegments();
      for (Segment segment : validSegments) {
        deleteDatamapData(segment);
      }
    } catch (IOException e) {
      LOGGER.error("drop datamap failed, failed to delete datamap directory");
    }
  }

  @Override
  public boolean willBecomeStale(TableOperation operation) {
    switch (operation) {
      case ALTER_RENAME:
        return false;
      case ALTER_DROP:
        return true;
      case ALTER_ADD_COLUMN:
        return false;
      case ALTER_CHANGE_DATATYPE:
        return true;
      case STREAMING:
        return false;
      case DELETE:
        return true;
      case UPDATE:
        return true;
      case PARTITION:
        return true;
      default:
        return false;
    }
  }

  @Override
  public boolean isOperationBlocked(TableOperation operation, Object... targets) {
    switch (operation) {
      case ALTER_DROP: {
        // alter table drop columns
        // will be blocked if the columns in bloomfilter datamap
        List<String> columnsToDrop = (List<String>) targets[0];
        List<String> indexedColumnNames = dataMapMeta.getIndexedColumnNames();
        for (String indexedcolumn : indexedColumnNames) {
          for (String column : columnsToDrop) {
            if (column.equalsIgnoreCase(indexedcolumn)) {
              return true;
            }
          }
        }
        return false;
      }
      case ALTER_CHANGE_DATATYPE: {
        // alter table change one column datatype
        // will be blocked if the column in bloomfilter datamap
        String columnToChangeDatatype = (String) targets[0];
        List<String> indexedColumnNames = dataMapMeta.getIndexedColumnNames();
        for (String indexedcolumn : indexedColumnNames) {
          if (indexedcolumn.equalsIgnoreCase(columnToChangeDatatype)) {
            return true;
          }
        }
        return false;
      }
      default:
        return false;
    }
  }

  @Override
  public DataMapMeta getMeta() {
    return this.dataMapMeta;
  }

  @Override
  public DataMapLevel getDataMapLevel() {
    return DataMapLevel.CG;
  }
}

1	/*
2	* Licensed to the Apache Software Foundation (ASF) under one or more
3	* contributor license agreements. See the NOTICE file distributed with
4	* this work for additional information regarding copyright ownership.
5	* The ASF licenses this file to You under the Apache License, Version 2.0
6	* (the "License"); you may not use this file except in compliance with
7	* the License. You may obtain a copy of the License at
8	*
9	* http://www.apache.org/licenses/LICENSE-2.0
10	*
11	* Unless required by applicable law or agreed to in writing, software
12	* distributed under the License is distributed on an "AS IS" BASIS,
13	* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14	* See the License for the specific language governing permissions and
15	* limitations under the License.
16	*/
17	package org.apache.carbondata.datamap.bloom;
18
19	import java.io.File;
20	import java.io.IOException;
21	import java.util.*;
22	import java.util.concurrent.ConcurrentHashMap;
23
24	import org.apache.carbondata.common.annotations.InterfaceAudience;
25	import org.apache.carbondata.common.exceptions.sql.MalformedDataMapCommandException;
26	import org.apache.carbondata.common.logging.LogService;
27	import org.apache.carbondata.common.logging.LogServiceFactory;
28	import org.apache.carbondata.core.cache.Cache;
29	import org.apache.carbondata.core.cache.CacheProvider;
30	import org.apache.carbondata.core.cache.CacheType;
31	import org.apache.carbondata.core.constants.CarbonV3DataFormatConstants;
32	import org.apache.carbondata.core.datamap.DataMapDistributable;
33	import org.apache.carbondata.core.datamap.DataMapLevel;
34	import org.apache.carbondata.core.datamap.DataMapMeta;
35	import org.apache.carbondata.core.datamap.Segment;
36	import org.apache.carbondata.core.datamap.dev.DataMapBuilder;
37	import org.apache.carbondata.core.datamap.dev.DataMapFactory;
38	import org.apache.carbondata.core.datamap.dev.DataMapWriter;
39	import org.apache.carbondata.core.datamap.dev.cgdatamap.CoarseGrainDataMap;
40	import org.apache.carbondata.core.datastore.block.SegmentProperties;
41	import org.apache.carbondata.core.datastore.filesystem.CarbonFile;
42	import org.apache.carbondata.core.datastore.impl.FileFactory;
43	import org.apache.carbondata.core.features.TableOperation;
44	import org.apache.carbondata.core.metadata.schema.table.CarbonTable;
45	import org.apache.carbondata.core.metadata.schema.table.DataMapSchema;
46	import org.apache.carbondata.core.metadata.schema.table.column.CarbonColumn;
47	import org.apache.carbondata.core.scan.filter.intf.ExpressionType;
48	import org.apache.carbondata.core.statusmanager.SegmentStatusManager;
49	import org.apache.carbondata.core.util.CarbonUtil;
50	import org.apache.carbondata.core.util.path.CarbonTablePath;
51	import org.apache.carbondata.events.Event;
52
53	import org.apache.commons.lang3.StringUtils;
54
55	/**
56	* This class is for Bloom Filter for blocklet level
57	*/
58	@InterfaceAudience.Internal
59	public class BloomCoarseGrainDataMapFactory extends DataMapFactory<CoarseGrainDataMap> {
60	private static final LogService LOGGER = LogServiceFactory.getLogService(	1✔
61	BloomCoarseGrainDataMapFactory.class.getName());	1✔
62	/**
63	* property for size of bloom filter
64	*/
65	private static final String BLOOM_SIZE = "bloom_size";
66	/**
67	* default size for bloom filter, cardinality of the column.
68	*/
69	private static final int DEFAULT_BLOOM_FILTER_SIZE =
70	CarbonV3DataFormatConstants.NUMBER_OF_ROWS_PER_BLOCKLET_COLUMN_PAGE_DEFAULT * 20;
71	/**
72	* property for fpp(false-positive-probability) of bloom filter
73	*/
74	private static final String BLOOM_FPP = "bloom_fpp";
75	/**
76	* default value for fpp of bloom filter is 0.001%
77	*/
78	private static final double DEFAULT_BLOOM_FILTER_FPP = 0.00001d;
79
80	/**
81	* property for compressing bloom while saving to disk.
82	*/
83	private static final String COMPRESS_BLOOM = "bloom_compress";
84	/**
85	* Default value of compressing bloom while save to disk.
86	*/
87	private static final boolean DEFAULT_BLOOM_COMPRESS = true;
88
89	private DataMapMeta dataMapMeta;
90	private String dataMapName;
91	private int bloomFilterSize;
92	private double bloomFilterFpp;
93	private boolean bloomCompress;
94	private Cache<BloomCacheKeyValue.CacheKey, BloomCacheKeyValue.CacheValue> cache;
95	// segmentId -> list of index file
96	private Map<String, Set<String>> segmentMap = new ConcurrentHashMap<>();	1✔
97
98	public BloomCoarseGrainDataMapFactory(CarbonTable carbonTable, DataMapSchema dataMapSchema)
99	throws MalformedDataMapCommandException {
100	super(carbonTable, dataMapSchema);	1✔
101	Objects.requireNonNull(carbonTable);	1✔
102	Objects.requireNonNull(dataMapSchema);	1✔
103
104	this.dataMapName = dataMapSchema.getDataMapName();	1✔
105
106	List<CarbonColumn> indexedColumns = carbonTable.getIndexedColumns(dataMapSchema);	1✔
107	this.bloomFilterSize = validateAndGetBloomFilterSize(dataMapSchema);	1✔
108	this.bloomFilterFpp = validateAndGetBloomFilterFpp(dataMapSchema);	1✔
109	this.bloomCompress = validateAndGetBloomCompress(dataMapSchema);	1✔
110	List<ExpressionType> optimizedOperations = new ArrayList<ExpressionType>();	1✔
111	// todo: support more optimize operations
112	optimizedOperations.add(ExpressionType.EQUALS);	1✔
113	optimizedOperations.add(ExpressionType.IN);	1✔
114	this.dataMapMeta = new DataMapMeta(this.dataMapName, indexedColumns, optimizedOperations);	1✔
115	LOGGER.info(String.format("DataMap %s works for %s with bloom size %d",	1✔
116	this.dataMapName, this.dataMapMeta, this.bloomFilterSize));	1✔
117	try {
118	this.cache = CacheProvider.getInstance()	1✔
119	.createCache(new CacheType("bloom_cache"), BloomDataMapCache.class.getName());	1✔
120	} catch (Exception e) {	×
121	LOGGER.error(e);	×
122	throw new MalformedDataMapCommandException(e.getMessage());	×
123	}	1✔
124	}	1✔
125
126	/**
127	* validate Lucene DataMap BLOOM_SIZE
128	* 1. BLOOM_SIZE property is optional, 32000 * 20 will be the default size.
129	* 2. BLOOM_SIZE should be an integer that greater than 0
130	*/
131	private int validateAndGetBloomFilterSize(DataMapSchema dmSchema)
132	throws MalformedDataMapCommandException {
133	String bloomFilterSizeStr = dmSchema.getProperties().get(BLOOM_SIZE);	1✔
134	if (StringUtils.isBlank(bloomFilterSizeStr)) {	1✔
135	LOGGER.warn(	1✔
136	String.format("Bloom filter size is not configured for datamap %s, use default value %d",	1✔
137	dataMapName, DEFAULT_BLOOM_FILTER_SIZE));	1✔
138	return DEFAULT_BLOOM_FILTER_SIZE;	1✔
139	}
140	int bloomFilterSize;
141	try {
142	bloomFilterSize = Integer.parseInt(bloomFilterSizeStr);	1✔
143	} catch (NumberFormatException e) {	×
144	throw new MalformedDataMapCommandException(	×
145	String.format("Invalid value of bloom filter size '%s', it should be an integer",	×
146	bloomFilterSizeStr));
147	}	1✔
148	// todo: reconsider the boundaries of bloom filter size
149	if (bloomFilterSize <= 0) {	1✔
150	throw new MalformedDataMapCommandException(	×
151	String.format("Invalid value of bloom filter size '%s', it should be greater than 0",	×
152	bloomFilterSizeStr));
153	}
154	return bloomFilterSize;	1✔
155	}
156
157	/**
158	* validate bloom DataMap BLOOM_FPP
159	* 1. BLOOM_FPP property is optional, 0.00001 will be the default value.
160	* 2. BLOOM_FPP should be (0, 1)
161	*/
162	private double validateAndGetBloomFilterFpp(DataMapSchema dmSchema)
163	throws MalformedDataMapCommandException {
164	String bloomFilterFppStr = dmSchema.getProperties().get(BLOOM_FPP);	1✔
165	if (StringUtils.isBlank(bloomFilterFppStr)) {	1✔
166	LOGGER.warn(	1✔
167	String.format("Bloom filter FPP is not configured for datamap %s, use default value %f",	1✔
168	dataMapName, DEFAULT_BLOOM_FILTER_FPP));	1✔
169	return DEFAULT_BLOOM_FILTER_FPP;	1✔
170	}
171	double bloomFilterFpp;
172	try {
173	bloomFilterFpp = Double.parseDouble(bloomFilterFppStr);	1✔
174	} catch (NumberFormatException e) {	×
175	throw new MalformedDataMapCommandException(	×
176	String.format("Invalid value of bloom filter fpp '%s', it should be an numeric",	×
177	bloomFilterFppStr));
178	}	1✔
179	if (bloomFilterFpp < 0 \|\| bloomFilterFpp - 1 >= 0) {	1✔
180	throw new MalformedDataMapCommandException(	×
181	String.format("Invalid value of bloom filter fpp '%s', it should be in range 0~1",	×
182	bloomFilterFppStr));
183	}
184	return bloomFilterFpp;	1✔
185	}
186
187	/**
188	* validate bloom DataMap COMPRESS_BLOOM
189	* Default value is true
190	*/
191	private boolean validateAndGetBloomCompress(DataMapSchema dmSchema) {
192	String bloomCompress = dmSchema.getProperties().get(COMPRESS_BLOOM);	1✔
193	if (StringUtils.isBlank(bloomCompress)) {	1✔
194	LOGGER.warn(	1✔
195	String.format("Bloom compress is not configured for datamap %s, use default value %b",	1✔
196	dataMapName, DEFAULT_BLOOM_COMPRESS));	1✔
197	return DEFAULT_BLOOM_COMPRESS;	1✔
198	}
199	return Boolean.parseBoolean(bloomCompress);	×
200	}
201
202	@Override
203	public DataMapWriter createWriter(Segment segment, String shardName,
204	SegmentProperties segmentProperties) throws IOException {
205	LOGGER.info(	1✔
206	String.format("Data of BloomCoarseGranDataMap %s for table %s will be written to %s",	1✔
207	this.dataMapName, getCarbonTable().getTableName() , shardName));	1✔
208	return new BloomDataMapWriter(getCarbonTable().getTablePath(), this.dataMapName,	1✔
209	this.dataMapMeta.getIndexedColumns(), segment, shardName, segmentProperties,	1✔
210	this.bloomFilterSize, this.bloomFilterFpp, bloomCompress);
211	}
212
213	@Override
214	public DataMapBuilder createBuilder(Segment segment, String shardName,
215	SegmentProperties segmentProperties) throws IOException {
216	return new BloomDataMapBuilder(getCarbonTable().getTablePath(), this.dataMapName,	1✔
217	this.dataMapMeta.getIndexedColumns(), segment, shardName, segmentProperties,	1✔
218	this.bloomFilterSize, this.bloomFilterFpp, bloomCompress);
219	}
220
221	/**
222	* returns all shard directories of bloom index files for query
223	* if bloom index files are merged we should get only one shard path
224	*/
225	private Set<String> getAllShardPaths(String tablePath, String segmentId) {
226	String dataMapStorePath = CarbonTablePath.getDataMapStorePath(	1✔
227	tablePath, segmentId, dataMapName);
228	CarbonFile[] carbonFiles = FileFactory.getCarbonFile(dataMapStorePath).listFiles();	1✔
229	Set<String> shardPaths = new HashSet<>();	1✔
230	boolean mergeShardInprogress = false;	1✔
231	CarbonFile mergeShardFile = null;	1✔
232	for (CarbonFile carbonFile : carbonFiles) {	1✔
233	if (carbonFile.getName().equals(BloomIndexFileStore.MERGE_BLOOM_INDEX_SHARD_NAME)) {	1✔
234	mergeShardFile = carbonFile;	1✔
235	} else if (carbonFile.getName().equals(BloomIndexFileStore.MERGE_INPROGRESS_FILE)) {	1✔
236	mergeShardInprogress = true;	×
237	} else if (carbonFile.isDirectory()) {	1✔
238	shardPaths.add(carbonFile.getAbsolutePath());	×
239	}
240	}
241	if (mergeShardFile != null && !mergeShardInprogress) {	1✔
242	// should only get one shard path if mergeShard is generated successfully
243	shardPaths.clear();	1✔
244	shardPaths.add(mergeShardFile.getAbsolutePath());	1✔
245	}
246	return shardPaths;	1✔
247	}
248
249	@Override
250	public List<CoarseGrainDataMap> getDataMaps(Segment segment) throws IOException {
251	List<CoarseGrainDataMap> dataMaps = new ArrayList<>();	1✔
252	try {
253	Set<String> shardPaths = segmentMap.get(segment.getSegmentNo());	1✔
254	if (shardPaths == null) {	1✔
255	shardPaths = getAllShardPaths(getCarbonTable().getTablePath(), segment.getSegmentNo());	1✔
256	segmentMap.put(segment.getSegmentNo(), shardPaths);	1✔
257	}
258	Set<String> filteredShards = segment.getFilteredIndexShardNames();	1✔
259	for (String shard : shardPaths) {	1✔
260	if (shard.endsWith(BloomIndexFileStore.MERGE_BLOOM_INDEX_SHARD_NAME) \|\|	1✔
261	filteredShards.contains(new File(shard).getName())) {	×
262	// Filter out the tasks which are filtered through Main datamap.
263	// for merge shard, shard pruning delay to be done before pruning blocklet
264	BloomCoarseGrainDataMap bloomDM = new BloomCoarseGrainDataMap();	1✔
265	bloomDM.init(new BloomDataMapModel(shard, cache, segment.getConfiguration()));	1✔
266	String dataMapStorePath = CarbonTablePath.getDataMapStorePath(	1✔
267	getCarbonTable().getTablePath(), segment.getSegmentNo(), dataMapName);	1✔
268	bloomDM.initIndexColumnConverters(getCarbonTable(), dataMapStorePath,	1✔
269	dataMapMeta.getIndexedColumns());	1✔
270	bloomDM.setFilteredShard(filteredShards);	1✔
271	dataMaps.add(bloomDM);	1✔
272	}
273	}	1✔
274	} catch (Exception e) {	×
275	throw new IOException("Error occurs while init Bloom DataMap", e);	×
276	}	1✔
277	return dataMaps;	1✔
278	}
279
280	@Override
281	public List<CoarseGrainDataMap> getDataMaps(DataMapDistributable distributable)
282	throws IOException {
283	List<CoarseGrainDataMap> dataMaps = new ArrayList<>();	1✔
284	String indexPath = ((BloomDataMapDistributable) distributable).getIndexPath();	1✔
285	Set<String> filteredShards = ((BloomDataMapDistributable) distributable).getFilteredShards();	1✔
286	BloomCoarseGrainDataMap bloomDM = new BloomCoarseGrainDataMap();	1✔
287	bloomDM.init(new BloomDataMapModel(indexPath, cache, FileFactory.getConfiguration()));	1✔
288	String dmStorePath = CarbonTablePath.getDataMapStorePath(	1✔
289	getCarbonTable().getTablePath(), distributable.getSegment().getSegmentNo(), dataMapName);	1✔
290	bloomDM.initIndexColumnConverters(getCarbonTable(), dmStorePath,	1✔
291	dataMapMeta.getIndexedColumns());	1✔
292	bloomDM.setFilteredShard(filteredShards);	1✔
293	dataMaps.add(bloomDM);	1✔
294	return dataMaps;	1✔
295	}
296
297
298	@Override
299	public List<DataMapDistributable> toDistributable(Segment segment) {
300	List<DataMapDistributable> dataMapDistributableList = new ArrayList<>();	1✔
301	Set<String> shardPaths = segmentMap.get(segment.getSegmentNo());	1✔
302	if (shardPaths == null) {	1✔
303	shardPaths = getAllShardPaths(getCarbonTable().getTablePath(), segment.getSegmentNo());	1✔
304	segmentMap.put(segment.getSegmentNo(), shardPaths);	1✔
305	}
306	Set<String> filteredShards = segment.getFilteredIndexShardNames();	1✔
307	for (String shardPath : shardPaths) {	1✔
308	// Filter out the tasks which are filtered through Main datamap.
309	// for merge shard, shard pruning delay to be done before pruning blocklet
310	if (shardPath.endsWith(BloomIndexFileStore.MERGE_BLOOM_INDEX_SHARD_NAME) \|\|	1✔
311	filteredShards.contains(new File(shardPath).getName())) {	×
312	DataMapDistributable bloomDataMapDistributable =	1✔
313	new BloomDataMapDistributable(shardPath, filteredShards);
314	dataMapDistributableList.add(bloomDataMapDistributable);	1✔
315	}
316	}	1✔
317	return dataMapDistributableList;	1✔
318	}
319
320	@Override
321	public void fireEvent(Event event) {
322
323	}	×
324
325	@Override
326	public void clear(Segment segment) {
327	Set<String> shards = segmentMap.remove(segment.getSegmentNo());	1✔
328	if (shards != null) {	1✔
329	for (String shard : shards) {	1✔
330	for (CarbonColumn carbonColumn : dataMapMeta.getIndexedColumns()) {	1✔
331	cache.invalidate(new BloomCacheKeyValue.CacheKey(shard, carbonColumn.getColName()));	1✔
332	}	1✔
333	}	1✔
334	}
335	}	1✔
336
337	@Override
338	public synchronized void clear() {
339	if (segmentMap.size() > 0) {	1✔
340	List<String> segments = new ArrayList<>(segmentMap.keySet());	1✔
341	for (String segmentId : segments) {	1✔
342	clear(new Segment(segmentId, null, null));	1✔
343	}	1✔
344	}
345	}	1✔
346
347	@Override
348	public void deleteDatamapData(Segment segment) throws IOException {
349	try {
350	String segmentId = segment.getSegmentNo();	1✔
351	String datamapPath = CarbonTablePath	1✔
352	.getDataMapStorePath(getCarbonTable().getTablePath(), segmentId, dataMapName);	1✔
353	if (FileFactory.isFileExist(datamapPath)) {	1✔
354	CarbonFile file = FileFactory.getCarbonFile(datamapPath,	1✔
355	FileFactory.getFileType(datamapPath));	1✔
356	CarbonUtil.deleteFoldersAndFilesSilent(file);	1✔
357	}
358	} catch (InterruptedException ex) {	×
359	throw new IOException("Failed to delete datamap for segment_" + segment.getSegmentNo());	×
360	}	1✔
361	}	1✔
362
363	@Override
364	public void deleteDatamapData() {
365	SegmentStatusManager ssm =	1✔
366	new SegmentStatusManager(getCarbonTable().getAbsoluteTableIdentifier());	1✔
367	try {
368	List<Segment> validSegments = ssm.getValidAndInvalidSegments().getValidSegments();	1✔
369	for (Segment segment : validSegments) {	1✔
370	deleteDatamapData(segment);	×
371	}	×
372	} catch (IOException e) {	×
373	LOGGER.error("drop datamap failed, failed to delete datamap directory");	×
374	}	1✔
375	}	1✔
376
377	@Override
378	public boolean willBecomeStale(TableOperation operation) {
379	switch (operation) {	1✔
380	case ALTER_RENAME:
381	return false;	1✔
382	case ALTER_DROP:
383	return true;	1✔
384	case ALTER_ADD_COLUMN:
385	return false;	1✔
386	case ALTER_CHANGE_DATATYPE:
387	return true;	1✔
388	case STREAMING:
389	return false;	×
390	case DELETE:
391	return true;	×
392	case UPDATE:
393	return true;	×
394	case PARTITION:
395	return true;	×
396	default:
397	return false;	×
398	}
399	}
400
401	@Override
402	public boolean isOperationBlocked(TableOperation operation, Object... targets) {
403	switch (operation) {	1✔
404	case ALTER_DROP: {
405	// alter table drop columns
406	// will be blocked if the columns in bloomfilter datamap
407	List<String> columnsToDrop = (List<String>) targets[0];	×
408	List<String> indexedColumnNames = dataMapMeta.getIndexedColumnNames();	×
409	for (String indexedcolumn : indexedColumnNames) {	×
410	for (String column : columnsToDrop) {	×
411	if (column.equalsIgnoreCase(indexedcolumn)) {	×
412	return true;	×
413	}
414	}	×
415	}	×
416	return false;	×
417	}
418	case ALTER_CHANGE_DATATYPE: {
419	// alter table change one column datatype
420	// will be blocked if the column in bloomfilter datamap
421	String columnToChangeDatatype = (String) targets[0];	×
422	List<String> indexedColumnNames = dataMapMeta.getIndexedColumnNames();	×
423	for (String indexedcolumn : indexedColumnNames) {	×
424	if (indexedcolumn.equalsIgnoreCase(columnToChangeDatatype)) {	×
425	return true;	×
426	}
427	}	×
428	return false;	×
429	}
430	default:
431	return false;	1✔
432	}
433	}
434
435	@Override
436	public DataMapMeta getMeta() {
437	return this.dataMapMeta;	1✔
438	}
439
440	@Override
441	public DataMapLevel getDataMapLevel() {
442	return DataMapLevel.CG;	1✔
443	}
444	}

apache / carbondata / 643

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous