[sis] 03/03: Share the netCDF coordinate axes accross netCDF files.

[de...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

desruisseaux pushed a commit to branch geoapi-4.0
in repository https://gitbox.apache.org/repos/asf/sis.git

commit 0574eb2e186b15e2bf6dbc6f6c017202c5367c7a
Author: Martin Desruisseaux <ma...@geomatys.com>
AuthorDate: Thu Dec 20 16:26:32 2018 +0100

    Share the netCDF coordinate axes accross netCDF files.
---
 .../main/java/org/apache/sis/internal/netcdf/Grid.java |  2 +-
 .../java/org/apache/sis/internal/netcdf/Variable.java  | 10 ++++++++++
 .../apache/sis/internal/netcdf/impl/VariableInfo.java  |  1 +
 .../sis/internal/netcdf/ucar/VariableWrapper.java      | 18 +++++++++++++++---
 4 files changed, 27 insertions(+), 4 deletions(-)

diff --git a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Grid.java b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Grid.java
index 606ec0b..1b74793 100644
--- a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Grid.java
+++ b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Grid.java
@@ -89,7 +89,7 @@ public abstract class Grid extends NamedElement {
     private boolean isCRSDetermined;
 
     /**
-     * The geometry of this grid (is extent, its CRS and its conversion to the CRS).
+     * The geometry of this grid (its extent, its CRS and its conversion to the CRS).
      * May be {@code null} even after we attempted to create it.
      */
     private GridGeometry geometry;
diff --git a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Variable.java b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Variable.java
index a440652..2d4b8ff 100644
--- a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Variable.java
+++ b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/Variable.java
@@ -32,6 +32,7 @@ import org.apache.sis.math.Vector;
 import org.apache.sis.math.DecimalFunctions;
 import org.apache.sis.measure.NumberRange;
 import org.apache.sis.util.Numbers;
+import org.apache.sis.util.collection.WeakHashSet;
 import org.apache.sis.util.logging.WarningListeners;
 import org.apache.sis.util.resources.Errors;
 import ucar.nc2.constants.CDM;                      // We use only String constants.
@@ -49,6 +50,15 @@ import ucar.nc2.constants.CF;
  */
 public abstract class Variable extends NamedElement {
     /**
+     * Pool of vectors created by the {@link #read()} method. This pool is used for sharing netCDF coordinate axes,
+     * since the same vectors tend to be repeated in many netCDF files produced by the same data producer. Because
+     * those vectors can be large, sharing common instances may save a lot of memory.
+     *
+     * <p>All shared vectors shall be considered read-only.</p>
+     */
+    protected static final WeakHashSet<Vector> SHARED_VECTORS = new WeakHashSet<>(Vector.class);
+
+    /**
      * Names of attributes where to fetch minimum and maximum sample values, in preference order.
      *
      * @see #getValidValues()
diff --git a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/impl/VariableInfo.java b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/impl/VariableInfo.java
index 99c9e82..03203a6 100644
--- a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/impl/VariableInfo.java
+++ b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/impl/VariableInfo.java
@@ -692,6 +692,7 @@ final class VariableInfo extends Variable implements Comparable<VariableInfo> {
             }
         }
         values = data.compress(tolerance);
+        values = SHARED_VECTORS.unique(values);
     }
 
     /**
diff --git a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/ucar/VariableWrapper.java b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/ucar/VariableWrapper.java
index b4d24cf..bed9c0c 100644
--- a/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/ucar/VariableWrapper.java
+++ b/storage/sis-netcdf/src/main/java/org/apache/sis/internal/netcdf/ucar/VariableWrapper.java
@@ -73,6 +73,13 @@ final class VariableWrapper extends Variable {
     private final VariableIF raw;
 
     /**
+     * The values of the whole variable, or {@code null} if not yet read. This vector should be assigned only
+     * for relatively small variables, or for variables that are critical to the use of other variables
+     * (for example the values in coordinate system axes).
+     */
+    private transient Vector values;
+
+    /**
      * Creates a new variable wrapping the given netCDF interface.
      */
     VariableWrapper(final WarningListeners<?> listeners, VariableIF v) {
@@ -359,12 +366,17 @@ final class VariableWrapper extends Variable {
     /**
      * Reads all the data for this variable and returns them as an array of a Java primitive type.
      * Multi-dimensional variables are flattened as a one-dimensional array (wrapped in a vector).
-     * This method may cache the returned vector, at UCAR library choice.
+     * This method caches the returned vector.
      */
     @Override
+    @SuppressWarnings("ReturnOfCollectionOrArrayField")
     public Vector read() throws IOException {
-        final Array array = variable.read();                // May be cached by the UCAR library.
-        return createDecimalVector(array.get1DJavaArray(array.getElementType()), variable.isUnsigned());
+        if (values == null) {
+            final Array array = variable.read();                // May be already cached by the UCAR library.
+            values = createDecimalVector(array.get1DJavaArray(array.getElementType()), variable.isUnsigned());
+            values = SHARED_VECTORS.unique(values);
+        }
+        return values;
     }
 
     /**