Spark Decode parquet

WritableColumnVector read Dict

sql/core/src/main/java/org/apache/spark/sql/execution/vectorized/WritableColumnVector.java

@Override
  public UTF8String getUTF8String(int rowId) {
    if (isNullAt(rowId)) return null;
    if (dictionary == null) {
      return arrayData().getBytesAsUTF8String(getArrayOffset(rowId), getArrayLength(rowId));
    } else {
      byte[] bytes = dictionary.decodeToBinary(dictionaryIds.getDictId(rowId));
      return UTF8String.fromBytes(bytes);
    }
  }

dictionaryIds 为对应的索引

  /**
   * Reusable column for ids of dictionary.
   */
  protected WritableColumnVector dictionaryIds;

Spark P