[BuildsOnTop Of Julien and Vinoo's work] [WIP] ALP pseudo decimal encoding

parquet-column/src/main/java/org/apache/parquet/column/Encoding.java

@@ -29,6 +29,8 @@
 import org.apache.parquet.bytes.BytesUtils;
 import org.apache.parquet.column.page.DictionaryPage;
 import org.apache.parquet.column.values.ValuesReader;
+import org.apache.parquet.column.values.alp.AlpValuesReaderForDouble;
+import org.apache.parquet.column.values.alp.AlpValuesReaderForFloat;
 import org.apache.parquet.column.values.bitpacking.ByteBitPackingValuesReader;
 import org.apache.parquet.column.values.bytestreamsplit.ByteStreamSplitValuesReaderForDouble;
 import org.apache.parquet.column.values.bytestreamsplit.ByteStreamSplitValuesReaderForFLBA;
@@ -147,6 +149,24 @@ public ValuesReader getValuesReader(ColumnDescriptor descriptor, ValuesType valu
     }
   },
 
+  /**
+   * Adaptive Lossless floating-Point (ALP) encoding for FLOAT and DOUBLE columns.
+   */
+  ALP {
+    @Override
+    public ValuesReader getValuesReader(ColumnDescriptor descriptor, ValuesType valuesType) {
+      switch (descriptor.getType()) {
+        case FLOAT:
+          return new AlpValuesReaderForFloat();
+        case DOUBLE:
+          return new AlpValuesReaderForDouble();
+        default:
+          throw new ParquetDecodingException(
+              "Encoding ALP is only supported for type FLOAT and DOUBLE, got " + descriptor.getType());
+      }
+    }
+  },
+
   /**
    * @deprecated This is no longer used, and has been replaced by {@link #RLE}
    * which is combination of bit packing and rle

parquet-column/src/main/java/org/apache/parquet/column/ParquetProperties.java

@@ -50,6 +50,7 @@ public class ParquetProperties {
   public static final int DEFAULT_DICTIONARY_PAGE_SIZE = DEFAULT_PAGE_SIZE;
   public static final boolean DEFAULT_IS_DICTIONARY_ENABLED = true;
   public static final boolean DEFAULT_IS_BYTE_STREAM_SPLIT_ENABLED = false;
+  public static final boolean DEFAULT_IS_ALP_ENABLED = false;
   public static final WriterVersion DEFAULT_WRITER_VERSION = WriterVersion.PARQUET_1_0;
   public static final boolean DEFAULT_ESTIMATE_ROW_COUNT_FOR_PAGE_SIZE_CHECK = true;
   public static final int DEFAULT_MINIMUM_RECORD_COUNT_FOR_CHECK = 100;
@@ -132,6 +133,7 @@ public static WriterVersion fromString(String name) {
   private final int pageRowCountLimit;
   private final boolean pageWriteChecksumEnabled;
   private final ColumnProperty<ByteStreamSplitMode> byteStreamSplitEnabled;
+  private final ColumnProperty<Boolean> alpEnabled;
   private final Map<String, String> extraMetaData;
   private final ColumnProperty<Boolean> statistics;
   private final ColumnProperty<Boolean> sizeStatistics;
@@ -164,6 +166,7 @@ private ParquetProperties(Builder builder) {
     this.pageRowCountLimit = builder.pageRowCountLimit;
     this.pageWriteChecksumEnabled = builder.pageWriteChecksumEnabled;
     this.byteStreamSplitEnabled = builder.byteStreamSplitEnabled.build();
+    this.alpEnabled = builder.alpEnabled.build();
     this.extraMetaData = builder.extraMetaData;
     this.statistics = builder.statistics.build();
     this.sizeStatistics = builder.sizeStatistics.build();
@@ -259,6 +262,20 @@ public boolean isByteStreamSplitEnabled(ColumnDescriptor column) {
     }
   }
 
+  /**
+   * Returns true if ALP encoding is enabled for the given column.
+   * ALP encoding is only applicable to FLOAT and DOUBLE columns.
+   */
+  public boolean isAlpEnabled(ColumnDescriptor column) {
+    switch (column.getPrimitiveType().getPrimitiveTypeName()) {
+      case FLOAT:
+      case DOUBLE:
+        return alpEnabled.getValue(column);
+      default:
+        return false;
+    }
+  }
+
   public ByteBufferAllocator getAllocator() {
     return allocator;
   }
@@ -416,6 +433,7 @@ public static class Builder {
     private int pageRowCountLimit = DEFAULT_PAGE_ROW_COUNT_LIMIT;
     private boolean pageWriteChecksumEnabled = DEFAULT_PAGE_WRITE_CHECKSUM_ENABLED;
     private final ColumnProperty.Builder<ByteStreamSplitMode> byteStreamSplitEnabled;
+    private final ColumnProperty.Builder<Boolean> alpEnabled;
     private Map<String, String> extraMetaData = new HashMap<>();
     private final ColumnProperty.Builder<Boolean> statistics;
     private final ColumnProperty.Builder<Boolean> sizeStatistics;
@@ -427,6 +445,7 @@ private Builder() {
               DEFAULT_IS_BYTE_STREAM_SPLIT_ENABLED
                   ? ByteStreamSplitMode.FLOATING_POINT
                   : ByteStreamSplitMode.NONE);
+      alpEnabled = ColumnProperty.<Boolean>builder().withDefaultValue(DEFAULT_IS_ALP_ENABLED);
       bloomFilterEnabled = ColumnProperty.<Boolean>builder().withDefaultValue(DEFAULT_BLOOM_FILTER_ENABLED);
       bloomFilterNDVs = ColumnProperty.<Long>builder().withDefaultValue(null);
       bloomFilterFPPs = ColumnProperty.<Double>builder().withDefaultValue(DEFAULT_BLOOM_FILTER_FPP);
@@ -457,6 +476,7 @@ private Builder(ParquetProperties toCopy) {
       this.numBloomFilterCandidates = ColumnProperty.builder(toCopy.numBloomFilterCandidates);
       this.maxBloomFilterBytes = toCopy.maxBloomFilterBytes;
       this.byteStreamSplitEnabled = ColumnProperty.builder(toCopy.byteStreamSplitEnabled);
+      this.alpEnabled = ColumnProperty.builder(toCopy.alpEnabled);
       this.extraMetaData = toCopy.extraMetaData;
       this.statistics = ColumnProperty.builder(toCopy.statistics);
       this.sizeStatistics = ColumnProperty.builder(toCopy.sizeStatistics);
@@ -534,6 +554,29 @@ public Builder withExtendedByteStreamSplitEncoding(boolean enable) {
       return this;
     }
 
+    /**
+     * Enable or disable ALP encoding for FLOAT and DOUBLE columns.
+     *
+     * @param enable whether ALP encoding should be enabled
+     * @return this builder for method chaining.
+     */
+    public Builder withAlpEncoding(boolean enable) {
+      this.alpEnabled.withDefaultValue(enable);
+      return this;
+    }
+
+    /**
+     * Enable or disable ALP encoding for the specified column.
+     *
+     * @param columnPath the path of the column (dot-string)
+     * @param enable     whether ALP encoding should be enabled
+     * @return this builder for method chaining.
+     */
+    public Builder withAlpEncoding(String columnPath, boolean enable) {
+      this.alpEnabled.withValue(columnPath, enable);
+      return this;
+    }
+
     /**
      * Set the Parquet format dictionary page size.
      *