Class DeltaByteArrayReader

java.lang.Object
org.apache.parquet.column.values.ValuesReader
org.apache.parquet.column.values.deltastrings.DeltaByteArrayReader
All Implemented Interfaces:
RequiresPreviousReader

public class DeltaByteArrayReader extends ValuesReader implements RequiresPreviousReader
Reads binary data written by DeltaByteArrayWriter
  • Constructor Details

    • DeltaByteArrayReader

      public DeltaByteArrayReader()
  • Method Details

    • initFromPage

      public void initFromPage(int valueCount, org.apache.parquet.bytes.ByteBufferInputStream stream) throws IOException
      Description copied from class: ValuesReader
      Called to initialize the column reader from a part of a page.

      Implementations must consume all bytes from the input stream, leaving the stream ready to read the next section of data. The underlying implementation knows how much data to read, so a length is not provided.

      Each page may contain several sections:

      • repetition levels column
      • definition levels column
      • data column
      Overrides:
      initFromPage in class ValuesReader
      Parameters:
      valueCount - count of values in this page
      stream - an input stream containing the page data at the correct offset
      Throws:
      IOException - if there is an exception while reading from the input stream
    • skip

      public void skip()
      Description copied from class: ValuesReader
      Skips the next value in the page
      Specified by:
      skip in class ValuesReader
    • readBytes

      public Binary readBytes()
      Overrides:
      readBytes in class ValuesReader
      Returns:
      the next Binary from the page
    • setPreviousReader

      public void setPreviousReader(ValuesReader reader)
      There was a bug (PARQUET-246) in which DeltaByteArrayWriter's reset() method did not clear the previous value state that it tracks internally. This resulted in the first value of all pages (except for the first page) to be a delta from the last value of the previous page. In order to read corrupted files written with this bug, when reading a new page we need to recover the previous page's last value to use it (if needed) to read the first value.
      Specified by:
      setPreviousReader in interface RequiresPreviousReader