Examples with VectorizedParquetRecordReader org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader used on opensource projects

Example 6 with VectorizedParquetRecordReader

use of org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader in project hive by apache.

the class TestVectorizedColumnReader method testNullSplitForParquetReader.

@Test
public void testNullSplitForParquetReader() throws Exception {
    Configuration conf = new Configuration();
    conf.set(IOConstants.COLUMNS, "int32_field");
    conf.set(IOConstants.COLUMNS_TYPES, "int");
    HiveConf.setBoolVar(conf, HiveConf.ConfVars.HIVE_VECTORIZATION_ENABLED, true);
    HiveConf.setVar(conf, HiveConf.ConfVars.PLAN, "//tmp");
    initialVectorizedRowBatchCtx(conf);
    VectorizedParquetRecordReader reader = new VectorizedParquetRecordReader((org.apache.hadoop.mapred.InputSplit) null, new JobConf(conf));
    assertFalse(reader.next(reader.createKey(), reader.createValue()));
}

Example 7 with VectorizedParquetRecordReader

use of org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader in project hive by apache.

the class VectorizedColumnReaderTestBase method structRead.

protected void structRead(boolean isDictionaryEncoding) throws Exception {
    Configuration conf = new Configuration();
    conf.set(IOConstants.COLUMNS, "struct_field");
    conf.set(IOConstants.COLUMNS_TYPES, "struct<a:int,b:double>");
    conf.setBoolean(ColumnProjectionUtils.READ_ALL_COLUMNS, false);
    conf.set(ColumnProjectionUtils.READ_COLUMN_IDS_CONF_STR, "0");
    String schema = "message hive_schema {\n" + "group struct_field {\n" + "  optional int32 a;\n" + "  optional double b;\n" + "}\n" + "}\n";
    VectorizedParquetRecordReader reader = createParquetReader(schema, conf);
    VectorizedRowBatch previous = reader.createValue();
    int c = 0;
    try {
        while (reader.next(NullWritable.get(), previous)) {
            StructColumnVector vector = (StructColumnVector) previous.cols[0];
            LongColumnVector cv = (LongColumnVector) vector.fields[0];
            DoubleColumnVector dv = (DoubleColumnVector) vector.fields[1];
            for (int i = 0; i < cv.vector.length; i++) {
                if (c == nElements) {
                    break;
                }
                assertEquals(getIntValue(isDictionaryEncoding, c), cv.vector[i]);
                assertEquals(getDoubleValue(isDictionaryEncoding, c), dv.vector[i], 0);
                assertFalse(vector.isNull[i]);
                assertFalse(vector.isRepeating);
                c++;
            }
        }
        assertEquals("It doesn't exit at expected position", nElements, c);
    } finally {
        reader.close();
    }
}

Example 8 with VectorizedParquetRecordReader

use of org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader in project hive by apache.

the class VectorizedColumnReaderTestBase method structReadSomeNull.

protected void structReadSomeNull(boolean isDictionaryEncoding) throws Exception {
    Configuration conf = new Configuration();
    conf.set(IOConstants.COLUMNS, "struct_field_some_null");
    conf.set(IOConstants.COLUMNS_TYPES, "struct<f:int,g:double>");
    conf.setBoolean(ColumnProjectionUtils.READ_ALL_COLUMNS, false);
    conf.set(ColumnProjectionUtils.READ_COLUMN_IDS_CONF_STR, "0");
    String schema = "message hive_schema {\n" + "group struct_field_some_null {\n" + "  optional int32 f;\n" + "  optional double g;\n" + "}\n";
    VectorizedParquetRecordReader reader = createParquetReader(schema, conf);
    VectorizedRowBatch previous = reader.createValue();
    int c = 0;
    try {
        while (reader.next(NullWritable.get(), previous)) {
            StructColumnVector sv = (StructColumnVector) previous.cols[0];
            LongColumnVector fv = (LongColumnVector) sv.fields[0];
            DoubleColumnVector gv = (DoubleColumnVector) sv.fields[1];
            for (int i = 0; i < fv.vector.length; i++) {
                if (c == nElements) {
                    break;
                }
                assertEquals(c % 2 == 0, fv.isNull[i]);
                assertEquals(c % 3 == 0, gv.isNull[i]);
                assertEquals(c % /* 2*3 = */
                6 == 0, sv.isNull[i]);
                if (!sv.isNull[i]) {
                    if (!fv.isNull[i]) {
                        assertEquals(getIntValue(isDictionaryEncoding, c), fv.vector[i]);
                    }
                    if (!gv.isNull[i]) {
                        assertEquals(getDoubleValue(isDictionaryEncoding, c), gv.vector[i], 0);
                    }
                }
                assertFalse(fv.isRepeating);
                c++;
            }
        }
        assertEquals("It doesn't exit at expected position", nElements, c);
    } finally {
        reader.close();
    }
}

Example 9 with VectorizedParquetRecordReader

use of org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader in project hive by apache.

the class VectorizedColumnReaderTestBase method longRead.

protected void longRead(boolean isDictionaryEncoding) throws Exception {
    Configuration conf = new Configuration();
    conf.set(IOConstants.COLUMNS, "int64_field");
    conf.set(IOConstants.COLUMNS_TYPES, "bigint");
    conf.setBoolean(ColumnProjectionUtils.READ_ALL_COLUMNS, false);
    conf.set(ColumnProjectionUtils.READ_COLUMN_IDS_CONF_STR, "0");
    VectorizedParquetRecordReader reader = createParquetReader("message test { required int64 int64_field;}", conf);
    VectorizedRowBatch previous = reader.createValue();
    try {
        int c = 0;
        while (reader.next(NullWritable.get(), previous)) {
            LongColumnVector vector = (LongColumnVector) previous.cols[0];
            assertTrue(vector.noNulls);
            for (int i = 0; i < vector.vector.length; i++) {
                if (c == nElements) {
                    break;
                }
                assertEquals("Failed at " + c, getLongValue(isDictionaryEncoding, c), vector.vector[i]);
                assertFalse(vector.isNull[i]);
                c++;
            }
        }
        assertEquals(nElements, c);
    } finally {
        reader.close();
    }
}

Example 10 with VectorizedParquetRecordReader

use of org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader in project hive by apache.

the class VectorizedColumnReaderTestBase method createParquetReader.

protected VectorizedParquetRecordReader createParquetReader(String schemaString, Configuration conf) throws IOException, InterruptedException, HiveException {
    conf.set(PARQUET_READ_SCHEMA, schemaString);
    HiveConf.setBoolVar(conf, HiveConf.ConfVars.HIVE_VECTORIZATION_ENABLED, true);
    HiveConf.setVar(conf, HiveConf.ConfVars.PLAN, "//tmp");
    Job vectorJob = new Job(conf, "read vector");
    ParquetInputFormat.setInputPaths(vectorJob, file);
    ParquetInputFormat parquetInputFormat = new ParquetInputFormat(GroupReadSupport.class);
    ParquetInputSplit split = (ParquetInputSplit) parquetInputFormat.getSplits(vectorJob).get(0);
    initialVectorizedRowBatchCtx(conf);
    return new VectorizedParquetRecordReader(split, new JobConf(conf));
}