More validation for partial read parameters

Merge pull request #116 from mathworks/partial-read-enh

Author: krisfed

Zarr.m

@@ -103,6 +103,13 @@ function pyReloadInProcess()
                     paramName)
             end
 
+            if any(params>dims)
+                error("MATLAB:Zarr:PartialReadOutOfBounds",...
+                    "Elements in %s must not exceed "+...
+                    "the corresponding Zarr array dimensions.",...
+                    paramName)
+            end
+
             newParams = params;
         end
 
@@ -315,20 +322,39 @@ function makeZarrGroups(existingParentPath, newGroupsPath)
             count = Zarr.processPartialReadParams(count, info.shape,...
                 maxCount, "Count"); 
 
+            if any(count>maxCount)
+                error("MATLAB:Zarr:PartialReadOutOfBounds",...
+                    "Requested Count in combination with other "+...
+                    "parameters exceeds Zarr array dimensions.")
+            end
+
             % Convert partial read parameters to tensorstore-style
             % indexing
             start = start - 1; % tensorstore is 0-based
             % Tensorstore uses end index instead of count
             % (it does NOT include element at the end index)
             endInds = start + stride.*count;
 
+            % Store the datatype
+            obj.Datatype = ZarrDatatype.fromZarrType(info.dtype);
+
+            % Check if reading requested data might exceed available memory
+            try
+                zeros(count, obj.Datatype.MATLABType);
+            catch ME
+                if strcmp(ME.identifier, 'MATLAB:array:SizeLimitExceeded')
+                    error("MATLAB:Zarr:OutOfMemory",...
+                        "Reading requested data (%s %s array) "+...
+                        "would exceed maximum array size preference. "+...
+                        "Select a smaller subset of data to read.",...
+                        join(string(count), "-by-"), obj.Datatype.MATLABType)
+                end
+            end
+
             % Read the data
             ndArrayData = Zarr.ZarrPy.readZarr(obj.KVStoreSchema,...
                 start, endInds, stride);
 
-            % Store the datatype
-            obj.Datatype = ZarrDatatype.fromTensorstoreType(ndArrayData.dtype.name);
-
             % Convert the numpy array to MATLAB array
             data = cast(ndArrayData, obj.Datatype.MATLABType);
         end

ZarrDatatype.m

@@ -74,6 +74,15 @@
             obj = ZarrDatatype(ind);
         end
 
+        function obj = fromZarrType(zarrType)
+            % Create a datatype object based on Zarr datatype name
+            arguments
+                zarrType (1,1) string {ZarrDatatype.mustBeZarrType}
+            end
+
+            ind = find(zarrType == ZarrDatatype.ZarrTypes);
+            obj = ZarrDatatype(ind);
+        end
 
         function mustBeMATLABType(type)
             % Validator for MATLAB types
@@ -84,6 +93,11 @@ function mustBeTensorstoreType(type)
             % Validator for Tensorstore types
             mustBeMember(type, ZarrDatatype.TensorstoreTypes)
         end
+
+        function mustBeZarrType(type)
+            % Validator for Zarr types
+            mustBeMember(type, ZarrDatatype.ZarrTypes)
+        end
     end
 
 end

test/tZarrRead.m

@@ -109,6 +109,16 @@ function nonExistentArray(testcase)
             testcase.verifyError(@()zarrread('nonexistent/'),errID);
         end
 
+        function tooBigArray(testcase)
+            % Verify zarrread error when a user tries to read data that is
+            % too large
+            
+            bigDataPath = "bigData/myzarr";
+            zarrcreate(bigDataPath, [100000,100000], Datatype='single');
+            errID = 'MATLAB:Zarr:OutOfMemory';
+            testcase.verifyError(@()zarrread(bigDataPath),errID);
+        end
+
         function invalidFilePath(testcase)
             % Verify zarrread error when an invalid file path is used.
 
@@ -147,7 +157,7 @@ function invalidPartialReadParams(testcase)
             testcase.verifyError(@()zarrread(zpath,Count=wrongDims),errID);
 
             % Invalid type
-            errID = 'MATLAB:validators:mustBeNumericOrLogical';
+            errID = 'MATLAB:validators:mustBeNumeric';
             testcase.verifyError(@()zarrread(zpath,"Start",""),errID);
             testcase.verifyError(@()zarrread(zpath,"Stride",""),errID);
             testcase.verifyError(@()zarrread(zpath,"Count",""),errID);
@@ -159,12 +169,17 @@ function invalidPartialReadParams(testcase)
             testcase.verifyError(@()zarrread(zpath,"Stride",inpVal),errID);
             testcase.verifyError(@()zarrread(zpath,"Count",inpVal),errID);
 
-            % Input out of bounds
+
+            % Parameters out of bounds
             inpVal = [100 200];
-            errID = 'MATLAB:Python:PyException';
+            errID = 'MATLAB:Zarr:PartialReadOutOfBounds';
             testcase.verifyError(@()zarrread(zpath,"Start",inpVal),errID);
-            %testcase.verifyError(@()zarrread(zpath,"Stride",inpVal),errID);
+            testcase.verifyError(@()zarrread(zpath,"Stride",inpVal),errID);
             testcase.verifyError(@()zarrread(zpath,"Count",inpVal),errID);
+
+            % Combination of parameters out of bounds
+            testcase.verifyError(...
+                @()zarrread(zpath,Start=[3 4],Count=[2 2]),errID)
         end
     end
 end

zarrread.m

@@ -27,9 +27,9 @@
 
 arguments
     filepath {mustBeTextScalar, mustBeNonzeroLengthText}
-    options.Start (1,:) {mustBeInteger, mustBePositive} = [];
-    options.Count (1,:) {mustBeInteger, mustBePositive} = [];
-    options.Stride (1,:) {mustBeInteger, mustBePositive} = [];
+    options.Start (1,:) {mustBeNumeric, mustBeInteger, mustBePositive} = [];
+    options.Count (1,:) {mustBeNumeric, mustBeInteger, mustBePositive} = [];
+    options.Stride (1,:) {mustBeNumeric, mustBeInteger, mustBePositive} = [];
 end
 
 zarrObj = Zarr(filepath);