add allowWraparound in bbox to ParquetProperties

Author: zhangfengcdt

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

@@ -65,7 +65,9 @@ public class ParquetProperties {
   public static final int DEFAULT_BLOOM_FILTER_CANDIDATES_NUMBER = 5;
   public static final boolean DEFAULT_STATISTICS_ENABLED = true;
   public static final boolean DEFAULT_SIZE_STATISTICS_ENABLED = true;
+  public static final boolean DEFAULT_GEO_STATISTICS_ENABLED = true;
   public static final boolean DEFAULT_GEOSPATIAL_STATISTICS_ENABLED = true;
+  public static final boolean DEFAULT_GEO_STATISTICS_WRAPAROUNDENABLED = true;
 
   public static final boolean DEFAULT_PAGE_WRITE_CHECKSUM_ENABLED = true;
 
@@ -116,6 +118,7 @@ public static WriterVersion fromString(String name) {
   private final boolean statisticsEnabled;
   private final boolean sizeStatisticsEnabled;
   private final boolean geospatialStatisticsEnabled;
+  private final boolean geospatialStatisticsBBoxWraparoundEnabled;
 
   // The expected NDV (number of distinct values) for each columns
   private final ColumnProperty<Long> bloomFilterNDVs;
@@ -131,6 +134,7 @@ public static WriterVersion fromString(String name) {
   private final ColumnProperty<Boolean> statistics;
   private final ColumnProperty<Boolean> sizeStatistics;
   private final ColumnProperty<Boolean> geospatialStatistics;
+  private final ColumnProperty<Boolean> geospatialStatisticsBBoxWraparound;
 
   private ParquetProperties(Builder builder) {
     this.pageSizeThreshold = builder.pageSize;
@@ -151,6 +155,7 @@ private ParquetProperties(Builder builder) {
     this.statisticsEnabled = builder.statisticsEnabled;
     this.sizeStatisticsEnabled = builder.sizeStatisticsEnabled;
     this.geospatialStatisticsEnabled = builder.geospatialStatisticsEnabled;
+    this.geospatialStatisticsBBoxWraparoundEnabled = builder.geospatialStatisticsBBoxWraparoundEnable;
     this.bloomFilterNDVs = builder.bloomFilterNDVs.build();
     this.bloomFilterFPPs = builder.bloomFilterFPPs.build();
     this.bloomFilterEnabled = builder.bloomFilterEnabled.build();
@@ -164,6 +169,7 @@ private ParquetProperties(Builder builder) {
     this.statistics = builder.statistics.build();
     this.sizeStatistics = builder.sizeStatistics.build();
     this.geospatialStatistics = builder.geospatialStatistics.build();
+    this.geospatialStatisticsBBoxWraparound = builder.geospatialStatisticsBBoxWraparound.build();
   }
 
   public static Builder builder() {
@@ -371,6 +377,10 @@ public boolean getGeoSpatialStatisticsEnabled(ColumnDescriptor column) {
     return geospatialStatisticsEnabled;
   }
 
+  public boolean getGeoSpatialStatisticsBBoxWraparoundEnabled(ColumnDescriptor column) {
+    return geospatialStatisticsBBoxWraparoundEnabled;
+  }
+
   @Override
   public String toString() {
     return "Parquet page size to " + getPageSizeThreshold() + '\n'
@@ -389,7 +399,8 @@ public String toString() {
         + "Page row count limit to " + getPageRowCountLimit() + '\n'
         + "Writing page checksums is: " + (getPageWriteChecksumEnabled() ? "on" : "off") + '\n'
         + "Statistics enabled: " + statisticsEnabled + '\n'
-        + "Size statistics enabled: " + sizeStatisticsEnabled;
+        + "Size statistics enabled: " + sizeStatisticsEnabled + '\n'
+        + "Geospatial statistics enabled: " + geospatialStatisticsEnabled;
   }
 
   public static class Builder {
@@ -407,7 +418,8 @@ public static class Builder {
     private int statisticsTruncateLength = DEFAULT_STATISTICS_TRUNCATE_LENGTH;
     private boolean statisticsEnabled = DEFAULT_STATISTICS_ENABLED;
     private boolean sizeStatisticsEnabled = DEFAULT_SIZE_STATISTICS_ENABLED;
-    private boolean geospatialStatisticsEnabled = DEFAULT_SIZE_STATISTICS_ENABLED;
+    private boolean geospatialStatisticsEnabled = DEFAULT_GEO_STATISTICS_ENABLED;
+    private boolean geospatialStatisticsBBoxWraparoundEnable = DEFAULT_GEO_STATISTICS_WRAPAROUNDENABLED;
     private final ColumnProperty.Builder<Long> bloomFilterNDVs;
     private final ColumnProperty.Builder<Double> bloomFilterFPPs;
     private int maxBloomFilterBytes = DEFAULT_MAX_BLOOM_FILTER_BYTES;
@@ -421,6 +433,7 @@ public static class Builder {
     private final ColumnProperty.Builder<Boolean> statistics;
     private final ColumnProperty.Builder<Boolean> sizeStatistics;
     private final ColumnProperty.Builder<Boolean> geospatialStatistics;
+    private final ColumnProperty.Builder<Boolean> geospatialStatisticsBBoxWraparound;
 
     private Builder() {
       enableDict = ColumnProperty.<Boolean>builder().withDefaultValue(DEFAULT_IS_DICTIONARY_ENABLED);
@@ -440,6 +453,8 @@ private Builder() {
       sizeStatistics = ColumnProperty.<Boolean>builder().withDefaultValue(DEFAULT_SIZE_STATISTICS_ENABLED);
       geospatialStatistics =
           ColumnProperty.<Boolean>builder().withDefaultValue(DEFAULT_GEOSPATIAL_STATISTICS_ENABLED);
+      geospatialStatisticsBBoxWraparound =
+          ColumnProperty.<Boolean>builder().withDefaultValue(DEFAULT_GEO_STATISTICS_WRAPAROUNDENABLED);
     }
 
     private Builder(ParquetProperties toCopy) {
@@ -465,6 +480,7 @@ private Builder(ParquetProperties toCopy) {
       this.statistics = ColumnProperty.builder(toCopy.statistics);
       this.sizeStatistics = ColumnProperty.builder(toCopy.sizeStatistics);
       this.geospatialStatistics = ColumnProperty.builder(toCopy.geospatialStatistics);
+      this.geospatialStatisticsBBoxWraparound = ColumnProperty.builder(toCopy.geospatialStatisticsBBoxWraparound);
     }
 
     /**

parquet-column/src/main/java/org/apache/parquet/column/impl/ColumnValueCollector.java

@@ -40,6 +40,7 @@ class ColumnValueCollector {
   private final boolean statisticsEnabled;
   private final boolean sizeStatisticsEnabled;
   private final boolean geospatialStatisticsEnabled;
+  private final boolean geospatialStatisticsBBoxWraparoundEnabled;
   private BloomFilterWriter bloomFilterWriter;
   private BloomFilter bloomFilter;
   private Statistics<?> statistics;
@@ -51,6 +52,7 @@ class ColumnValueCollector {
     this.statisticsEnabled = props.getStatisticsEnabled(path);
     this.sizeStatisticsEnabled = props.getSizeStatisticsEnabled(path);
     this.geospatialStatisticsEnabled = props.getGeoSpatialStatisticsEnabled(path);
+    this.geospatialStatisticsBBoxWraparoundEnabled = props.getGeoSpatialStatisticsBBoxWraparoundEnabled(path);
     resetPageStatistics();
     initBloomFilter(bloomFilterWriter, props);
   }
@@ -65,7 +67,7 @@ void resetPageStatistics() {
         : SizeStatistics.noopBuilder(
             path.getPrimitiveType(), path.getMaxRepetitionLevel(), path.getMaxDefinitionLevel());
     this.geospatialStatisticsBuilder = geospatialStatisticsEnabled
-        ? GeospatialStatistics.newBuilder(path.getPrimitiveType())
+        ? GeospatialStatistics.newBuilder(path.getPrimitiveType(), geospatialStatisticsBBoxWraparoundEnabled)
         : GeospatialStatistics.noopBuilder();
   }
 

parquet-column/src/main/java/org/apache/parquet/column/statistics/geometry/BoundingBox.java

@@ -25,7 +25,7 @@
 
 public class BoundingBox {
 
-  boolean allowWraparound = Boolean.parseBoolean(System.getenv().getOrDefault("ALLOW_BBOX_WRAPAROUND", "true"));
+  private boolean allowWraparound = true;
 
   private double xMin = Double.POSITIVE_INFINITY;
   private double xMax = Double.NEGATIVE_INFINITY;
@@ -50,6 +50,10 @@ public BoundingBox(
 
   public BoundingBox() {}
 
+  void enableWraparound(boolean enable) {
+    allowWraparound = enable;
+  }
+
   public double getXMin() {
     return xMin;
   }
@@ -216,7 +220,8 @@ public BoundingBox copy() {
 
   @Override
   public String toString() {
-    return "BoundingBox{" + "xMin="
+    return "BoundingBox{" + "allowWraparound="
+        + allowWraparound + ", xMin="
         + xMin + ", xMax="
         + xMax + ", yMin="
         + yMin + ", yMax="

parquet-column/src/main/java/org/apache/parquet/column/statistics/geometry/GeospatialStatistics.java

@@ -155,6 +155,15 @@ public static GeospatialStatistics.Builder newBuilder(PrimitiveType type) {
     }
   }
 
+  /** Create a new GeospatialStatistics builder with the additional geospatialStatisticsBBoxWraparoundEnabled property */
+  public static Builder newBuilder(PrimitiveType primitiveType, boolean geospatialStatisticsBBoxWraparoundEnabled) {
+    Builder builder = newBuilder(primitiveType);
+    if (geospatialStatisticsBBoxWraparoundEnabled) {
+      builder.boundingBox.enableWraparound(true);
+    }
+    return builder;
+  }
+
   /**
    * Constructs a GeospatialStatistics object with the specified CRS, bounding box, and geospatial types.
    *