Package parquet.schema

Examples of parquet.schema.MessageType

33
34
35
36
37
38
39
40
import parquet.schema.Type.Repetition;

public class HiveSchemaConverter {

  public static MessageType convert(final List<String> columnNames, final List<TypeInfo> columnTypes) {
    final MessageType schema = new MessageType("hive_schema", convertTypes(columnNames, columnTypes));
    return schema;
  }
View Full Code Here
36
37
38
39
40
41
42
43
import parquet.schema.Types;

public class HiveSchemaConverter {

  public static MessageType convert(final List<String> columnNames, final List<TypeInfo> columnTypes) {
    final MessageType schema = new MessageType("hive_schema", convertTypes(columnNames, columnTypes));
    return schema;
  }
View Full Code Here
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
            // below allows schema evolution
            typeListTable.add(new PrimitiveType(Repetition.OPTIONAL, PrimitiveTypeName.BINARY, col));
          }
        }
      }
      MessageType tableSchema = new MessageType(TABLE_SCHEMA, typeListTable);
      contextMetadata.put(HIVE_SCHEMA_KEY, tableSchema.toString());

      final List<Integer> indexColumnsWanted = ColumnProjectionUtils.getReadColumnIDs(configuration);

      final List<Type> typeListWanted = new ArrayList<Type>();

      for (final Integer idx : indexColumnsWanted) {
        if (idx < listColumns.size()) {
          String col = listColumns.get(idx);
          if (indexAccess) {
              typeListWanted.add(fileSchema.getFields().get(idx));
          } else {
            col = col.toLowerCase();
            if (lowerCaseFileSchemaColumns.containsKey(col)) {
              typeListWanted.add(tableSchema.getType(lowerCaseFileSchemaColumns.get(col)));
            }
          }
        }
      }
      MessageType requestedSchemaByUser = new MessageType(fileSchema.getName(), typeListWanted);
      return new ReadContext(requestedSchemaByUser, contextMetadata);
    } else {
      contextMetadata.put(HIVE_SCHEMA_KEY, fileSchema.toString());
      return new ReadContext(fileSchema, contextMetadata);
    }
View Full Code Here
151
152
153
154
155
156
157
158
    final Map<String, String> metadata = readContext.getReadSupportMetadata();
    if (metadata == null) {
      throw new IllegalStateException("ReadContext not initialized properly. " +
        "Don't know the Hive Schema.");
    }
    final MessageType tableSchema = MessageTypeParser.parseMessageType(metadata.get(HIVE_SCHEMA_KEY));
    return new DataWritableRecordConverter(readContext.getRequestedSchema(), tableSchema);
  }
View Full Code Here
58
59
60
61
62
63
64
65
66
  }

  private void testConversion(final String columnNamesStr, final String columnsTypeStr, final String expectedSchema) throws Exception {
    final List<String> columnNames = createHiveColumnsFrom(columnNamesStr);
    final List<TypeInfo> columnTypes = createHiveTypeInfoFrom(columnsTypeStr);
    final MessageType messageTypeFound = HiveSchemaConverter.convert(columnNames, columnTypes);
    final MessageType expectedMT = MessageTypeParser.parseMessageType(expectedSchema);
    assertEquals("converting " + columnNamesStr + ": " + columnsTypeStr + " to " + expectedSchema, expectedMT, messageTypeFound);
  }
View Full Code Here
179
180
181
182
183
184
185
186
187
188
189
190
191
192
  public void testMapOriginalType() throws Exception {
    final String hiveColumnTypes = "map<string,string>";
    final String hiveColumnNames = "mapCol";
    final List<String> columnNames = createHiveColumnsFrom(hiveColumnNames);
    final List<TypeInfo> columnTypes = createHiveTypeInfoFrom(hiveColumnTypes);
    final MessageType messageTypeFound = HiveSchemaConverter.convert(columnNames, columnTypes);
    // this messageType only has one optional field, whose name is mapCol, original Type is MAP
    assertEquals(1, messageTypeFound.getFieldCount());
    parquet.schema.Type topLevel = messageTypeFound.getFields().get(0);
    assertEquals("mapCol",topLevel.getName());
    assertEquals(OriginalType.MAP, topLevel.getOriginalType());
    assertEquals(Repetition.OPTIONAL, topLevel.getRepetition());

    assertEquals(1, topLevel.asGroupType().getFieldCount());
View Full Code Here
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
    // remove the last semicolon, java really needs a join method for strings...
    // TODO - nvm apparently it requires a semicolon after every field decl, might want to file a bug
    //messageSchema = messageSchema.substring(schemaType, messageSchema.length() - 1);
    messageSchema += "}";

    MessageType schema = MessageTypeParser.parseMessageType(messageSchema);

    CompressionCodecName codec = CompressionCodecName.UNCOMPRESSED;
    ParquetFileWriter w = new ParquetFileWriter(configuration, schema, path);
    w.start();
    HashMap<String, Integer> columnValuesWritten = new HashMap();
    int valsWritten;
    for (int k = 0; k < props.numberRowGroups; k++){
      w.startBlock(props.recordsPerRowGroup);
      currentBooleanByte = 0;
      booleanBitCounter.reset();

      for (FieldInfo fieldInfo : props.fields.values()) {

        if ( ! columnValuesWritten.containsKey(fieldInfo.name)){
          columnValuesWritten.put((String) fieldInfo.name, 0);
          valsWritten = 0;
        } else {
          valsWritten = columnValuesWritten.get(fieldInfo.name);
        }

        String[] path1 = {(String) fieldInfo.name};
        ColumnDescriptor c1 = schema.getColumnDescription(path1);

        w.startColumn(c1, props.recordsPerRowGroup, codec);
        int valsPerPage = (int) Math.ceil(props.recordsPerRowGroup / (float) fieldInfo.numberOfPages);
        byte[] bytes;
        RunLengthBitPackingHybridValuesWriter defLevels = new RunLengthBitPackingHybridValuesWriter(
View Full Code Here
211
212
213
214
215
216
217
218
219
220
221
222
223
224
      if (blocksForCurrentSplit.size() == 0) {
        LOG.debug("HDFS block without row group: " + hdfsBlocks[i]);
      } else {
        long length = 0;
        for (BlockMetaData block : blocksForCurrentSplit) {
          MessageType requested = MessageTypeParser.parseMessageType(requestedSchema);
          List<ColumnChunkMetaData> columns = block.getColumns();
          for (ColumnChunkMetaData column : columns) {
            if (requested.containsPath(column.getPath().toArray())) {
              length += column.getTotalSize();
            }
          }
        }
        splits.add(new ParquetInputSplit(
View Full Code Here
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
  }

  private static void add(ParquetMetadata footer) {
    for (BlockMetaData blockMetaData : footer.getBlocks()) {
      ++ blockCount;
      MessageType schema = footer.getFileMetaData().getSchema();
      recordCount += blockMetaData.getRowCount();
      List<ColumnChunkMetaData> columns = blockMetaData.getColumns();
      for (ColumnChunkMetaData columnMetaData : columns) {
        ColumnDescriptor desc = schema.getColumnDescription(columnMetaData.getPath().toArray());
        add(
            desc,
            columnMetaData.getValueCount(),
            columnMetaData.getTotalSize(),
            columnMetaData.getTotalUncompressedSize(),
View Full Code Here
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
   * @return the result of the merge
   */
  static GlobalMetaData mergeInto(
      FileMetaData toMerge,
      GlobalMetaData mergedMetadata) {
    MessageType schema = null;
    Map<String, Set<String>> newKeyValues = new HashMap<String, Set<String>>();
    Set<String> createdBy = new HashSet<String>();
    if (mergedMetadata != null) {
      schema = mergedMetadata.getSchema();
      newKeyValues.putAll(mergedMetadata.getKeyValueMetaData());
      createdBy.addAll(mergedMetadata.getCreatedBy());
    }
    if ((schema == null && toMerge.getSchema() != null)
        || (schema != null && !schema.equals(toMerge.getSchema()))) {
      schema = mergeInto(toMerge.getSchema(), schema);
    }
    for (Entry<String, String> entry : toMerge.getKeyValueMetaData().entrySet()) {
      Set<String> values = newKeyValues.get(entry.getKey());
      if (values == null) {
View Full Code Here
TOP

Related Classes of parquet.schema.MessageType

  • com.asakusafw.directio.hive.parquet.DataModelMaterializer
  • com.asakusafw.directio.hive.parquet.DataModelWriteSupport
  • com.facebook.presto.hive.ParquetHiveRecordCursor$PrestoReadSupport
  • org.apache.drill.exec.store.parquet.ParquetRecordWriter
  • org.apache.drill.exec.store.parquet.ParquetScanBatchCreator
  • org.apache.drill.exec.store.parquet.TestFileGenerator
  • org.apache.drill.exec.store.parquet2.DrillParquetReader
  • org.apache.drill.exec.store.ParquetRecordReaderTest
  • org.apache.drill.ParquetSchemaMerge
  • org.apache.hadoop.hive.ql.io.parquet.convert.HiveSchemaConverter
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.