Zarr improvements

cdm/core/src/main/java/ucar/nc2/constants/CDM.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1998-2018 John Caron and University Corporation for Atmospheric Research/Unidata
+ * Copyright (c) 1998-2025 John Caron and University Corporation for Atmospheric Research/Unidata
  * See LICENSE for license information.
  */
 
@@ -66,6 +66,7 @@ public class CDM {
   public static final String NCPROPERTIES = "_NCProperties";
   public static final String ISNETCDF4 = "_IsNetcdf4";
   public static final String SUPERBLOCKVERSION = "_SuperblockVersion";
+  public static final String ARRAYDIMENSIONS = "_ARRAY_DIMENSIONS";
 
   public static final String[] SPECIALS = {NCPROPERTIES, ISNETCDF4, SUPERBLOCKVERSION};
 

cdm/zarr/src/main/java/ucar/nc2/iosp/zarr/ZarrHeader.java

@@ -1,10 +1,12 @@
 /*
- * Copyright (c) 2021 University Corporation for Atmospheric Research/Unidata
+ * Copyright (c) 2021-2025 University Corporation for Atmospheric Research/Unidata
  * See LICENSE for license information.
  */
 
 package ucar.nc2.iosp.zarr;
 
+import static ucar.nc2.constants.CDM.ARRAYDIMENSIONS;
+
 import com.fasterxml.jackson.databind.ObjectMapper;
 
 import ucar.ma2.ArrayObject;
@@ -34,6 +36,7 @@ public class ZarrHeader {
   private final RandomAccessDirectory rootRaf;
   private final Group.Builder rootGroup;
   private final String rootLocation;
+
   private static final ObjectMapper objectMapper = new ObjectMapper();
 
   public ZarrHeader(RandomAccessDirectory raf, Group.Builder rootGroup) {
@@ -219,20 +222,25 @@ private void makeVariable(RandomAccessDirectoryItem item, long dataOffset, ZArra
 
       for (Attribute attr : attrs) {
         final String attrName = attr.getName();
-        if ("_ARRAY_DIMENSIONS".equals(attrName)) {
+        if (ARRAYDIMENSIONS.equals(attrName)) {
           try {
-            final ArrayObject.D1 aod1 = (ArrayObject.D1) attr.getValues();
-
-            // getSize returns a long
-            final int aodSize = (int) aod1.getSize();
-            dimNames = new String[aodSize];
-
-            for (int i = 0; i < aodSize; ++i) {
-              dimNames[i] = (String) aod1.get(i);
+            if (attr.getLength() == 1 && attr.getStringValue().equals("")) {
+              // scalar array without a named dimension
+              logger.debug("  {} is a scalar array without a named dimension", vname);
+            } else {
+              final ArrayObject.D1 aod1 = (ArrayObject.D1) attr.getValues();
+
+              // getSize returns a long
+              final int aodSize = (int) aod1.getSize();
+              dimNames = new String[aodSize];
+
+              for (int i = 0; i < aodSize; ++i) {
+                dimNames[i] = (String) aod1.get(i);
+              }
+              hasNamedDimensions = true;
             }
-            hasNamedDimensions = true;
           } catch (final Exception exc) {
-            logger.debug("  Could not extract _ARRAY_DIMENSIONS for {}, {}", vname, exc.getMessage());
+            logger.debug("  Could not extract {} for {}, {}", ARRAYDIMENSIONS, vname, exc.getMessage());
           }
         }
       }
@@ -260,7 +268,8 @@ private void makeVariable(RandomAccessDirectoryItem item, long dataOffset, ZArra
       final Dimension.Builder dim = Dimension.builder(dname, shape[i]);
       dim.setIsVariableLength(false);
       dim.setIsUnlimited(false);
-      dim.setIsShared(false);
+      // if using named dimensions from _ARRAY_DIMENSIONS, mark the dimension as shared
+      dim.setIsShared(hasNamedDimensions);
 
       final Dimension dd = dim.build();
 
@@ -275,6 +284,9 @@ private void makeVariable(RandomAccessDirectoryItem item, long dataOffset, ZArra
           if (dd.getLength() != prevd.getLength()) {
             throw new ZarrFormatException("Named dimension " + dname + " seen with inconsistent lengths.");
           }
+          // replace newly created dimension with the previously added dimension
+          dims.remove(dd);
+          dims.add(prevd);
         } else {
           logger.trace("adding {} to group as a shared dimension", dname);
           parentGroup.addDimension(dd);
@@ -327,7 +339,13 @@ private List<Attribute> makeAttributes(RandomAccessDirectoryItem item) {
         Attribute.Builder attr = Attribute.builder(key);
         Object val = attrMap.get(key);
         if (val instanceof Collection<?>) {
-          attr.setValues(Arrays.asList(((Collection) val).toArray()), false);
+          Collection<?> collection = (Collection<?>) val;
+          if (collection.isEmpty() && key.equals(ARRAYDIMENSIONS)) {
+            // scalar array
+            attr.setValues(Collections.singletonList(""), false);
+          } else {
+            attr.setValues(Arrays.asList(collection.toArray()), false);
+          }
         } else if (val instanceof Number) {
           attr.setNumericValue((Number) val, false);
         } else {
@@ -354,7 +372,8 @@ private static int getChunkIndex(RandomAccessDirectoryItem item, ZArray zarray)
 
     int nDims = zarray.getShape().length;
     // verify is data file, else return -1
-    String pattern = String.format("([0-9]+%c){%d}[0-9]+", zarray.getSeparator().charAt(0), nDims - 1);
+    String pattern = String.format("([0-9]+%c){%d}[0-9]+", zarray.getSeparator().charAt(0), nDims == 0 ? 0 : nDims - 1);
+
     if (!fileName.matches(pattern)) {
       return -1;
     }
@@ -363,14 +382,20 @@ private static int getChunkIndex(RandomAccessDirectoryItem item, ZArray zarray)
     String[] dims = fileName.split(String.format("\\%c", zarray.getSeparator().charAt(0)));
     int[] subs = Arrays.stream(dims).mapToInt(dim -> Integer.parseInt(dim)).toArray();
 
-    // get number of chunks in each dimension
-    int[] nChunks = new int[nDims];
-    int[] shape = zarray.getShape();
-    int[] chunkSize = zarray.getChunks();
-    for (int i = 0; i < nDims; i++) {
-      nChunks[i] = (int) Math.ceil(shape[i] / chunkSize[i]);
+    // find chunk number as a flat index
+    if (nDims != 0) {
+      // get number of chunks in each dimension
+      int[] nChunks = new int[nDims];
+      int[] shape = zarray.getShape();
+      int[] chunkSize = zarray.getChunks();
+      for (int i = 0; i < nDims; i++) {
+        nChunks[i] = (int) Math.ceil(shape[i] / chunkSize[i]);
+      }
+      return ZarrUtils.subscriptsToIndex(subs, nChunks);
+    } else {
+      // scalar array
+      return 0;
     }
-    return ZarrUtils.subscriptsToIndex(subs, nChunks);
   }
 
   /**

cdm/zarr/src/main/java/ucar/nc2/iosp/zarr/ZarrLayoutBB.java

@@ -1,3 +1,8 @@
+/*
+ * Copyright (c) 2021-2025 University Corporation for Atmospheric Research/Unidata
+ * See LICENSE for license information.
+ */
+
 package ucar.nc2.iosp.zarr;
 
 import ucar.ma2.Range;
@@ -127,14 +132,20 @@ public LayoutBBTiled.DataChunk next() {
 
     private void incrementChunk() {
       // increment index from inner dimension outward
-      int i = this.currChunk.length - 1;
-      while (this.currChunk[i] + 1 >= nChunks[i] && i > 0) {
-        this.currChunk[i] = 0;
-        i--;
+      if (this.currChunk.length != 0) {
+        int i = this.currChunk.length - 1;
+        while (this.currChunk[i] + 1 >= nChunks[i] && i > 0) {
+          this.currChunk[i] = 0;
+          i--;
+        }
+        this.currChunk[i]++;
+        this.currOffset += initializedChunks.getOrDefault(this.chunkNum, (long) 0);
+        this.chunkNum = ZarrUtils.subscriptsToIndex(this.currChunk, nChunks);
+      } else {
+        // scalar array
+        this.currOffset = 0;
+        this.chunkNum = 0;
       }
-      this.currChunk[i]++;
-      this.currOffset += initializedChunks.getOrDefault(this.chunkNum, (long) 0);
-      this.chunkNum = ZarrUtils.subscriptsToIndex(this.currChunk, nChunks);
     }
   }
 

cdm/zarr/src/test/data/geozarr/README

@@ -0,0 +1,5 @@
+Examples obtained from https://github.com/zarr-developers/geozarr-spec/tree/cnl-examples/ on 2025-04-28 (commit d1eabecbea63d6f300600d529d7818bbf6df5bcc).
+No changes were made.
+
+Examples are licensed under Creative Commons Attribution 4.0 International.
+Full text can be found at https://raw.githubusercontent.com/zarr-developers/geozarr-spec/refs/heads/cnl-examples/LICENSE

cdm/zarr/src/test/data/geozarr/xyt-raster.zarr/.zattrs

@@ -0,0 +1,7 @@
+{
+    "Conventions": "CF-1.8",
+    "profile": [
+        "time-series-raster",
+        "scalar-raster"
+    ]
+}

cdm/zarr/src/test/data/geozarr/xyt-raster.zarr/.zgroup

@@ -0,0 +1,3 @@
+{
+    "zarr_format": 2
+}

cdm/zarr/src/test/data/geozarr/xyt-raster.zarr/.zmetadata

@@ -0,0 +1,164 @@
+{
+    "metadata": {
+        ".zattrs": {
+            "Conventions": "CF-1.8",
+            "profile": [
+                "time-series-raster",
+                "scalar-raster"
+            ]
+        },
+        ".zgroup": {
+            "zarr_format": 2
+        },
+        "spatial_ref/.zarray": {
+            "chunks": [],
+            "compressor": null,
+            "dtype": "<i8",
+            "fill_value": null,
+            "filters": null,
+            "order": "C",
+            "shape": [],
+            "zarr_format": 2
+        },
+        "spatial_ref/.zattrs": {
+            "GeoTransform": "319281.23 12.5 0 5639331.75 0 -12.5",
+            "_ARRAY_DIMENSIONS": [],
+            "crs_wkt": "PROJCS[\"WGS 84 / UTM zone 30N\",GEOGCS[\"WGS 84\",DATUM[\"WGS_1984\",SPHEROID[\"WGS 84\",6378137,298.257223563,AUTHORITY[\"EPSG\",\"7030\"]],AUTHORITY[\"EPSG\",\"6326\"]],PRIMEM[\"Greenwich\",0],UNIT[\"degree\",0.0174532925199433,AUTHORITY[\"EPSG\",\"9122\"]],AUTHORITY[\"EPSG\",\"4326\"]],PROJECTION[\"Transverse_Mercator\"],PARAMETER[\"latitude_of_origin\",0],PARAMETER[\"central_meridian\",-3],PARAMETER[\"scale_factor\",0.9996],PARAMETER[\"false_easting\",500000],PARAMETER[\"false_northing\",0],UNIT[\"metre\",1,AUTHORITY[\"EPSG\",\"9001\"]],AXIS[\"Easting\",EAST],AXIS[\"Northing\",NORTH],AUTHORITY[\"EPSG\",\"32630\"]]",
+            "false_easting": 500000.0,
+            "false_northing": 0.0,
+            "geographic_crs_name": "WGS 84",
+            "grid_mapping_name": "transverse_mercator",
+            "horizontal_datum_name": "World Geodetic System 1984",
+            "inverse_flattening": 298.257223563,
+            "latitude_of_projection_origin": 0.0,
+            "longitude_of_central_meridian": -3.0,
+            "prime_meridian_name": "Greenwich",
+            "projected_crs_name": "WGS 84 / UTM zone 30N",
+            "reference_ellipsoid_name": "WGS 84",
+            "scale_factor_at_central_meridian": 0.9996,
+            "semi_major_axis": 6378137.0,
+            "semi_minor_axis": 6356752.314245179
+        },
+        "temperature/.zarray": {
+            "chunks": [
+                2,
+                128,
+                128
+            ],
+            "compressor": {
+                "blocksize": 0,
+                "clevel": 5,
+                "cname": "lz4",
+                "id": "blosc",
+                "shuffle": 1
+            },
+            "dtype": "<f8",
+            "fill_value": "NaN",
+            "filters": null,
+            "order": "C",
+            "shape": [
+                4,
+                128,
+                128
+            ],
+            "zarr_format": 2
+        },
+        "temperature/.zattrs": {
+            "AREA_OR_POINT": "Point",
+            "_ARRAY_DIMENSIONS": [
+                "time",
+                "y",
+                "x"
+            ],
+            "grid_mapping": "spatial_ref",
+            "standard_name": "air_temperature",
+            "units": "K"
+        },
+        "time/.zarray": {
+            "chunks": [
+                4
+            ],
+            "compressor": {
+                "blocksize": 0,
+                "clevel": 5,
+                "cname": "lz4",
+                "id": "blosc",
+                "shuffle": 1
+            },
+            "dtype": "<i8",
+            "fill_value": null,
+            "filters": null,
+            "order": "C",
+            "shape": [
+                4
+            ],
+            "zarr_format": 2
+        },
+        "time/.zattrs": {
+            "_ARRAY_DIMENSIONS": [
+                "time"
+            ],
+            "calendar": "proleptic_gregorian",
+            "long_name": "time of observation",
+            "standard_name": "time",
+            "units": "days since 2020-01-01 00:00:00"
+        },
+        "x/.zarray": {
+            "chunks": [
+                128
+            ],
+            "compressor": {
+                "blocksize": 0,
+                "clevel": 5,
+                "cname": "lz4",
+                "id": "blosc",
+                "shuffle": 1
+            },
+            "dtype": "<f8",
+            "fill_value": "NaN",
+            "filters": null,
+            "order": "C",
+            "shape": [
+                128
+            ],
+            "zarr_format": 2
+        },
+        "x/.zattrs": {
+            "_ARRAY_DIMENSIONS": [
+                "x"
+            ],
+            "long_name": "x coordinate of projection",
+            "standard_name": "projection_x_coordinate",
+            "units": "metre"
+        },
+        "y/.zarray": {
+            "chunks": [
+                128
+            ],
+            "compressor": {
+                "blocksize": 0,
+                "clevel": 5,
+                "cname": "lz4",
+                "id": "blosc",
+                "shuffle": 1
+            },
+            "dtype": "<f8",
+            "fill_value": "NaN",
+            "filters": null,
+            "order": "C",
+            "shape": [
+                128
+            ],
+            "zarr_format": 2
+        },
+        "y/.zattrs": {
+            "_ARRAY_DIMENSIONS": [
+                "y"
+            ],
+            "long_name": "y coordinate of projection",
+            "standard_name": "projection_y_coordinate",
+            "units": "metre"
+        }
+    },
+    "zarr_consolidated_format": 1
+}

cdm/zarr/src/test/data/geozarr/xyt-raster.zarr/spatial_ref/.zarray

@@ -0,0 +1,10 @@
+{
+    "chunks": [],
+    "compressor": null,
+    "dtype": "<i8",
+    "fill_value": null,
+    "filters": null,
+    "order": "C",
+    "shape": [],
+    "zarr_format": 2
+}