Package parquet.column.values.rle

Examples of parquet.column.values.rle.RunLengthBitPackingHybridValuesWriter

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
        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(MAX_EXPECTED_BIT_WIDTH_FOR_DEFINITION_LEVELS, valsPerPage);
        RunLengthBitPackingHybridValuesWriter repLevels = new RunLengthBitPackingHybridValuesWriter(MAX_EXPECTED_BIT_WIDTH_FOR_DEFINITION_LEVELS, valsPerPage);
        // for variable length binary fields
        int bytesNeededToEncodeLength = 4;
        if ((int) fieldInfo.bitLength > 0) {
          bytes = new byte[(int) Math.ceil(valsPerPage * (int) fieldInfo.bitLength / 8.0)];
        } else {
          // the twelve at the end is to account for storing a 4 byte length with each value
          int totalValLength = ((byte[]) fieldInfo.values[0]).length + ((byte[]) fieldInfo.values[1]).length + ((byte[]) fieldInfo.values[2]).length + 3 * bytesNeededToEncodeLength;
          // used for the case where there is a number of values in this row group that is not divisible by 3
          int leftOverBytes = 0;
          if ( valsPerPage % 3 > 0 ) leftOverBytes += ((byte[])fieldInfo.values[1]).length + bytesNeededToEncodeLength;
          if ( valsPerPage % 3 > 1 ) leftOverBytes += ((byte[])fieldInfo.values[2]).length + bytesNeededToEncodeLength;
          bytes = new byte[valsPerPage / 3 * totalValLength + leftOverBytes];
        }
        int bytesPerPage = (int) (valsPerPage * ((int) fieldInfo.bitLength / 8.0));
        int bytesWritten = 0;
        for (int z = 0; z < (int) fieldInfo.numberOfPages; z++, bytesWritten = 0) {
          for (int i = 0; i < valsPerPage; i++) {
            repLevels.writeInteger(0);
            defLevels.writeInteger(1);
            //System.out.print(i + ", " + (i % 25 == 0 ? "\n gen " + fieldInfo.name + ": " : ""));
            if (fieldInfo.values[0] instanceof Boolean) {

              bytes[currentBooleanByte] |= bitFields[booleanBitCounter.val]
                  & ((boolean) fieldInfo.values[valsWritten % 3] ? allBitsTrue : allBitsFalse);
              booleanBitCounter.increment();
              if (booleanBitCounter.val == 0) {
                currentBooleanByte++;
              }
              valsWritten++;
              if (currentBooleanByte > bytesPerPage) break;
            } else {
              if (fieldInfo.values[valsWritten % 3] instanceof byte[]){
                System.arraycopy(ByteArrayUtil.toByta(((byte[])fieldInfo.values[valsWritten % 3]).length),
                    0, bytes, bytesWritten, bytesNeededToEncodeLength);
                System.arraycopy(fieldInfo.values[valsWritten % 3],
                    0, bytes, bytesWritten + bytesNeededToEncodeLength, ((byte[])fieldInfo.values[valsWritten % 3]).length);
                bytesWritten += ((byte[])fieldInfo.values[valsWritten % 3]).length + bytesNeededToEncodeLength;
              } else{
                System.arraycopy( ByteArrayUtil.toByta(fieldInfo.values[valsWritten % 3]),
                    0, bytes, i * ((int) fieldInfo.bitLength / 8), (int) fieldInfo.bitLength / 8);
              }
              valsWritten++;
            }

          }
          byte[] fullPage = new byte[2 * 4 * valsPerPage + bytes.length];
          byte[] repLevelBytes = repLevels.getBytes().toByteArray();
          byte[] defLevelBytes = defLevels.getBytes().toByteArray();
          System.arraycopy(bytes, 0, fullPage, 0, bytes.length);
          System.arraycopy(repLevelBytes, 0, fullPage, bytes.length, repLevelBytes.length);
          System.arraycopy(defLevelBytes, 0, fullPage, bytes.length + repLevelBytes.length, defLevelBytes.length);
          w.writeDataPage( (props.recordsPerRowGroup / fieldInfo.numberOfPages), fullPage.length, BytesInput.from(fullPage), RLE, RLE, PLAIN);
View Full Code Here
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
        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(
          MAX_EXPECTED_BIT_WIDTH_FOR_DEFINITION_LEVELS,
          valsPerPage,
          new DirectByteBufferAllocator());
        RunLengthBitPackingHybridValuesWriter repLevels = new RunLengthBitPackingHybridValuesWriter(
          MAX_EXPECTED_BIT_WIDTH_FOR_DEFINITION_LEVELS,
          valsPerPage,
          new DirectByteBufferAllocator());
        // for variable length binary fields
        int bytesNeededToEncodeLength = 4;
        if ((int) fieldInfo.bitLength > 0) {
          bytes = new byte[(int) Math.ceil(valsPerPage * (int) fieldInfo.bitLength / 8.0)];
        } else {
          // the twelve at the end is to account for storing a 4 byte length with each value
          int totalValLength = ((byte[]) fieldInfo.values[0]).length + ((byte[]) fieldInfo.values[1]).length + ((byte[]) fieldInfo.values[2]).length + 3 * bytesNeededToEncodeLength;
          // used for the case where there is a number of values in this row group that is not divisible by 3
          int leftOverBytes = 0;
          if ( valsPerPage % 3 > 0 ) leftOverBytes += ((byte[])fieldInfo.values[1]).length + bytesNeededToEncodeLength;
          if ( valsPerPage % 3 > 1 ) leftOverBytes += ((byte[])fieldInfo.values[2]).length + bytesNeededToEncodeLength;
          bytes = new byte[valsPerPage / 3 * totalValLength + leftOverBytes];
        }
        int bytesPerPage = (int) (valsPerPage * ((int) fieldInfo.bitLength / 8.0));
        int bytesWritten = 0;
        for (int z = 0; z < (int) fieldInfo.numberOfPages; z++, bytesWritten = 0) {
          for (int i = 0; i < valsPerPage; i++) {
            repLevels.writeInteger(0);
            defLevels.writeInteger(1);
            //System.out.print(i + ", " + (i % 25 == 0 ? "\n gen " + fieldInfo.name + ": " : ""));
            if (fieldInfo.values[0] instanceof Boolean) {

              bytes[currentBooleanByte] |= bitFields[booleanBitCounter.val]
                  & ((boolean) fieldInfo.values[valsWritten % 3] ? allBitsTrue : allBitsFalse);
              booleanBitCounter.increment();
              if (booleanBitCounter.val == 0) {
                currentBooleanByte++;
              }
              valsWritten++;
              if (currentBooleanByte > bytesPerPage) break;
            } else {
              if (fieldInfo.values[valsWritten % 3] instanceof byte[]){
                System.arraycopy(ByteArrayUtil.toByta(((byte[])fieldInfo.values[valsWritten % 3]).length),
                    0, bytes, bytesWritten, bytesNeededToEncodeLength);
                System.arraycopy(fieldInfo.values[valsWritten % 3],
                    0, bytes, bytesWritten + bytesNeededToEncodeLength, ((byte[])fieldInfo.values[valsWritten % 3]).length);
                bytesWritten += ((byte[])fieldInfo.values[valsWritten % 3]).length + bytesNeededToEncodeLength;
              } else{
                System.arraycopy( ByteArrayUtil.toByta(fieldInfo.values[valsWritten % 3]),
                    0, bytes, i * ((int) fieldInfo.bitLength / 8), (int) fieldInfo.bitLength / 8);
              }
              valsWritten++;
            }

          }
          byte[] fullPage = new byte[2 * 4 * valsPerPage + bytes.length];
          byte[] repLevelBytes = repLevels.getBytes().toByteArray();
          byte[] defLevelBytes = defLevels.getBytes().toByteArray();
          System.arraycopy(bytes, 0, fullPage, 0, bytes.length);
          System.arraycopy(repLevelBytes, 0, fullPage, bytes.length, repLevelBytes.length);
          System.arraycopy(defLevelBytes, 0, fullPage, bytes.length + repLevelBytes.length, defLevelBytes.length);
          w.writeDataPage( (props.recordsPerRowGroup / fieldInfo.numberOfPages), fullPage.length, BytesInput.from(fullPage), RLE, RLE, PLAIN);
View Full Code Here
55
56
57
58
59
60
61
62
63
 
  public static ValuesWriter getColumnDescriptorValuesWriter(int maxLevel,  int initialSizePerCol) {
    if (maxLevel == 0) {
      return new DevNullValuesWriter();
    } else {
      return new RunLengthBitPackingHybridValuesWriter(
          BytesUtils.getWidthFromMaxInt(maxLevel), initialSizePerCol);
    }
  }
View Full Code Here
66
67
68
69
70
71
72
73
74
75
76
    switch (path.getType()) {
    case BOOLEAN:
      if(writerVersion == WriterVersion.PARQUET_1_0) {
        return new BooleanPlainValuesWriter();
      } else if (writerVersion == WriterVersion.PARQUET_2_0) {
        return new RunLengthBitPackingHybridValuesWriter(1, initialSizePerCol);
      }
      break;
    case BINARY:
      if(enableDictionary) {
        return new PlainBinaryDictionaryValuesWriter(dictionaryPageSizeThreshold, initialSizePerCol);
View Full Code Here
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
        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(
          MAX_EXPECTED_BIT_WIDTH_FOR_DEFINITION_LEVELS,
          valsPerPage,
          new DirectByteBufferAllocator());
        RunLengthBitPackingHybridValuesWriter repLevels = new RunLengthBitPackingHybridValuesWriter(
          MAX_EXPECTED_BIT_WIDTH_FOR_DEFINITION_LEVELS,
          valsPerPage,
          new DirectByteBufferAllocator());
        // for variable length binary fields
        int bytesNeededToEncodeLength = 4;
        if ((int) fieldInfo.bitLength > 0) {
          bytes = new byte[(int) Math.ceil(valsPerPage * (int) fieldInfo.bitLength / 8.0)];
        } else {
          // the twelve at the end is to account for storing a 4 byte length with each value
          int totalValLength = ((byte[]) fieldInfo.values[0]).length + ((byte[]) fieldInfo.values[1]).length + ((byte[]) fieldInfo.values[2]).length + 3 * bytesNeededToEncodeLength;
          // used for the case where there is a number of values in this row group that is not divisible by 3
          int leftOverBytes = 0;
          if ( valsPerPage % 3 > 0 ) {
            leftOverBytes += ((byte[])fieldInfo.values[1]).length + bytesNeededToEncodeLength;
          }
          if ( valsPerPage % 3 > 1 ) {
            leftOverBytes += ((byte[])fieldInfo.values[2]).length + bytesNeededToEncodeLength;
          }
          bytes = new byte[valsPerPage / 3 * totalValLength + leftOverBytes];
        }
        int bytesPerPage = (int) (valsPerPage * ((int) fieldInfo.bitLength / 8.0));
        int bytesWritten = 0;
        for (int z = 0; z < (int) fieldInfo.numberOfPages; z++, bytesWritten = 0) {
          for (int i = 0; i < valsPerPage; i++) {
            repLevels.writeInteger(0);
            defLevels.writeInteger(1);
            //System.out.print(i + ", " + (i % 25 == 0 ? "\n gen " + fieldInfo.name + ": " : ""));
            if (fieldInfo.values[0] instanceof Boolean) {

              bytes[currentBooleanByte] |= bitFields[booleanBitCounter.val]
                  & ((boolean) fieldInfo.values[valsWritten % 3] ? allBitsTrue : allBitsFalse);
              booleanBitCounter.increment();
              if (booleanBitCounter.val == 0) {
                currentBooleanByte++;
              }
              valsWritten++;
              if (currentBooleanByte > bytesPerPage) {
                break;
              }
            } else {
              if (fieldInfo.values[valsWritten % 3] instanceof byte[]) {
                System.arraycopy(ByteArrayUtil.toByta(((byte[])fieldInfo.values[valsWritten % 3]).length),
                    0, bytes, bytesWritten, bytesNeededToEncodeLength);
                System.arraycopy(fieldInfo.values[valsWritten % 3],
                    0, bytes, bytesWritten + bytesNeededToEncodeLength, ((byte[])fieldInfo.values[valsWritten % 3]).length);
                bytesWritten += ((byte[])fieldInfo.values[valsWritten % 3]).length + bytesNeededToEncodeLength;
              } else{
                System.arraycopy( ByteArrayUtil.toByta(fieldInfo.values[valsWritten % 3]),
                    0, bytes, i * ((int) fieldInfo.bitLength / 8), (int) fieldInfo.bitLength / 8);
              }
              valsWritten++;
            }

          }
          byte[] fullPage = new byte[2 * 4 * valsPerPage + bytes.length];
          byte[] repLevelBytes = repLevels.getBytes().toByteArray();
          byte[] defLevelBytes = defLevels.getBytes().toByteArray();
          System.arraycopy(bytes, 0, fullPage, 0, bytes.length);
          System.arraycopy(repLevelBytes, 0, fullPage, bytes.length, repLevelBytes.length);
          System.arraycopy(defLevelBytes, 0, fullPage, bytes.length + repLevelBytes.length, defLevelBytes.length);
          w.writeDataPage( (props.recordsPerRowGroup / fieldInfo.numberOfPages), fullPage.length, BytesInput.from(fullPage), RLE, RLE, PLAIN);
View Full Code Here
TOP

Related Classes of parquet.column.values.rle.RunLengthBitPackingHybridValuesWriter

  • org.apache.drill.exec.store.parquet.TestFileGenerator
  • parquet.bytes.BytesInput
  • parquet.column.ParquetProperties
  • java.io.ByteArrayOutputStream
  • parquet.io.ParquetEncodingException
    • 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.