HDF5: string/char read/write scalar/array

Author: scivision

+stdlib/+hdf5nc/h5save.m

@@ -70,7 +70,7 @@ function new_file(filename, varname, A, sizeA)
 end
 
 if isempty(sizeA)
-  if isscalar(A) && ~isstring(A)
+  if isscalar(A)
     h5_write_scalar(filename, varname, A)
   elseif isvector(A)
     h5create(filename, varname, length(A), 'DataType', class(A))
@@ -79,7 +79,7 @@ function new_file(filename, varname, A, sizeA)
   end
 else
   if isscalar(sizeA)
-    if sizeA == 0 && ~isstring(A)
+    if sizeA == 0
       h5_write_scalar(filename, varname, A)
     else
       h5create(filename, varname, sizeA, 'DataType', class(A))

+stdlib/+hdf5nc/private/coerce_ds.m

@@ -33,10 +33,10 @@
      if ~isa(A, 'int64')
        A = int64(A);
      end
-  case {'char', 'string'}
-    if ~isstring(A)
-      A = string(A);
-    end
+  case 'char'
+    A = string(A);
+  case 'string'
+    % pass
   otherwise, error('create_ds:type_error', 'unknown data type %s', dtype)
 end
 

+stdlib/+hdf5nc/private/h5_write_scalar.m

@@ -12,9 +12,20 @@ function h5_write_scalar(filename, varname, A)
 create_hdf5_group(fid, varname);
 
 space_id = H5S.create('H5S_SCALAR');
-t = class2h5t(A);
-type_id = H5T.copy(t);
+if isstring(A)
+  A = char(A);
+end
+if ischar(A)
+  type_id = H5T.copy('H5T_C_S1');
+  H5T.set_cset(type_id, H5ML.get_constant_value('H5T_CSET_UTF8'));
+  H5T.set_size(type_id, 'H5T_VARIABLE');
+  H5T.set_strpad(type_id, 'H5T_STR_NULLTERM');
+else
+  type_id = H5T.copy(class2h5t(A));
+end
+
 dset_id = H5D.create(fid, varname, type_id, space_id, dcpl);
+
 H5D.write(dset_id,'H5ML_DEFAULT','H5S_ALL','H5S_ALL', dcpl, A);
 
 H5S.close(space_id);

+stdlib/TestHDF5.m

@@ -23,12 +23,16 @@ function setup_file(tc)
 A3 = A2(:,1:3,1);
 A3(:,:,2) = 2*A3;
 A4(:,:,:,5) = A3;
+utf = 'Hello There 😄';
+utf2 = [utf; "☎"];
 
 tc.TestData.A0 = A0;
 tc.TestData.A1 = A1;
 tc.TestData.A2 = A2;
 tc.TestData.A3 = A3;
 tc.TestData.A4 = A4;
+tc.TestData.utf = utf;
+tc.TestData.utf2 = utf2;
 
 basic = tempname + ".h5";
 tc.TestData.basic = basic;
@@ -39,6 +43,8 @@ function setup_file(tc)
 h5save(basic, '/A2', A2)
 h5save(basic, '/A3', A3, "size", size(A3))
 h5save(basic, '/A4', A4)
+h5save(basic, "/utf", utf)
+h5save(basic, "/utf2", utf2)
 
 h5save(basic, '/t/x', 12)
 h5save(basic, '/t/y', 13)
@@ -73,7 +79,7 @@ function test_get_variables(tc)
 basic = tc.TestData.basic;
 
 v = h5variables(basic);
-tc.verifyEqual(sort(v), ["A0", "A1", "A2", "A3", "A4"])
+tc.verifyEqual(sort(v), ["A0", "A1", "A2", "A3", "A4", "utf", "utf2"])
 
 [v1,g] = h5variables(basic);
 tc.verifyEqual(v,v1)
@@ -145,6 +151,15 @@ function test_size(tc)
 tc.verifyEqual(s, [4,3,2,5])
 tc.verifyEqual(r, 4)
 
+r = h5ndims(basic, '/utf');
+s = h5size(basic, '/utf');
+tc.verifyEmpty(s)
+tc.verifyEqual(r, 0)
+
+r = h5ndims(basic, '/utf2');
+s = h5size(basic, '/utf2');
+tc.verifyEqual(s, 2)
+tc.verifyEqual(r, 1)
 end
 
 
@@ -153,7 +168,7 @@ function test_read(tc)
 
 s = h5read(basic, '/A0');
 tc.verifyTrue(isscalar(s))
-tc.verifyEqual(s, 42)
+tc.verifyEqual(s, tc.TestData.A0)
 
 s = h5read(basic, '/A1');
 tc.verifyTrue(isvector(s))
@@ -170,6 +185,15 @@ function test_read(tc)
 s = h5read(basic, '/A4');
 tc.verifyEqual(ndims(s), 4)
 tc.verifyEqual(s, tc.TestData.A4)
+
+s = h5read(basic, '/utf');
+tc.verifyTrue(ischar(s))
+tc.verifyEqual(s, tc.TestData.utf)
+
+s = h5read(basic, '/utf2');
+tc.verifyTrue(isstring(s))
+tc.verifyEqual(s, tc.TestData.utf2)
+
 end
 
 
@@ -222,7 +246,7 @@ function test_string(tc, str)
 h5save(basic, "/"+str, str)
 
 a = h5read(basic, "/"+str);
-tc.verifyEqual(a, string(str))
+tc.verifyEqual(a, char(str))
 end
 
 function test_name_only(tc)