Support file I/O for string columns in SolutionArray

interfaces/cython/cantera/composite.py

@@ -447,6 +447,7 @@ class SolutionArray:
         # From Transport
         'viscosity', 'electrical_conductivity', 'thermal_conductivity',
     ]
+    _strings = ['phase_of_matter']
     _n_species = [
         # from ThermoPhase
         'Y', 'X', 'concentrations', 'partial_molar_enthalpies',
@@ -955,14 +956,22 @@ def read_csv(self, filename):
         using `restore_data`. This method allows for recreation of data
         previously exported by `write_csv`.
         """
-        # read data block and header separately
-        data = np.genfromtxt(filename, skip_header=1, delimiter=',')
-        labels = np.genfromtxt(filename, max_rows=1, delimiter=',', dtype=str)
-
-        data_dict = OrderedDict()
-        for i, label in enumerate(labels):
-            data_dict[label] = data[:, i]
-
+        if np.lib.NumpyVersion(np.__version__) < "1.14.0":
+            # bytestring needs to be converted for columns containing strings
+            data = np.genfromtxt(filename, delimiter=',',
+                                 dtype=None, names=True)
+            data_dict = OrderedDict()
+            for label in data.dtype.names:
+                if data[label].dtype.type == np.bytes_:
+                    data_dict[label] = data[label].astype('U')
+                else:
+                    data_dict[label] = data[label]
+        else:
+            # the 'encoding' parameter introduced with NumPy 1.14 simplifies import
+            data = np.genfromtxt(filename, delimiter=',',
+                                 dtype=None, names=True, encoding=None)
+            data_dict = OrderedDict({label: data[label]
+                                     for label in data.dtype.names})
         self.restore_data(data_dict)
 
     def to_pandas(self, cols=None, *args, **kwargs):
@@ -1115,8 +1124,12 @@ def write_hdf(self, filename, *args, cols=None, group=None, subgroup=None,
             # store SolutionArray data
             for key, val in self._meta.items():
                 dgroup.attrs[key] = val
-            for header, col in data.items():
-                dgroup.create_dataset(header, data=col, **hdf_kwargs)
+            for header, value in data.items():
+                if value.dtype.type == np.str_:
+                    dgroup.create_dataset(header, data=value.astype('S'),
+                                          **hdf_kwargs)
+                else:
+                    dgroup.create_dataset(header, data=value, **hdf_kwargs)
 
         return group
 
@@ -1204,7 +1217,11 @@ def strip_ext(source):
             # load data
             data = OrderedDict()
             for name, value in dgroup.items():
-                if name != 'phase':
+                if name == 'phase':
+                    continue
+                elif value.dtype.type == np.bytes_:
+                    data[name] = np.array(value).astype('U')
+                else:
                     data[name] = np.array(value)
 
         self.restore_data(data)
@@ -1306,6 +1323,12 @@ def _make_functions():
     def empty_scalar(self):
         return np.empty(self._shape)
 
+    def empty_strings(self):
+        # The maximum length of strings assigned by built-in methods is
+        # currently limited to 50 characters; an attempt to assign longer
+        # character arrays will result in truncated strings.
+        return np.empty(self._shape, dtype='U50')
+
     def empty_species(self):
         return np.empty(self._shape + (self._phase.n_selected_species,))
 
@@ -1332,6 +1355,9 @@ def getter(self):
     for name in SolutionArray._scalar:
         setattr(SolutionArray, name, make_prop(name, empty_scalar, Solution))
 
+    for name in SolutionArray._strings:
+        setattr(SolutionArray, name, make_prop(name, empty_strings, Solution))
+
     for name in SolutionArray._n_species:
         setattr(SolutionArray, name, make_prop(name, empty_species, Solution))
 

interfaces/cython/cantera/test/test_composite.py

@@ -160,6 +160,16 @@ def test_write_csv(self):
         self.assertTrue(np.allclose(states.P, b.P))
         self.assertTrue(np.allclose(states.X, b.X))
 
+    def test_write_csv_str_column(self):
+        states = ct.SolutionArray(self.gas, 3, extra={'spam': 'eggs'})
+
+        outfile = pjoin(self.test_work_dir, 'solutionarray.csv')
+        states.write_csv(outfile)
+
+        b = ct.SolutionArray(self.gas, extra={'spam'})
+        b.read_csv(outfile)
+        self.assertEqual(list(states.spam), list(b.spam))
+
     @utilities.unittest.skipIf(isinstance(_pandas, ImportError), "pandas is not installed")
     def test_to_pandas(self):
 
@@ -219,6 +229,17 @@ def test_write_hdf(self):
         c.read_hdf(outfile, group='foo/bar/baz')
         self.assertTrue(np.allclose(states.T, c.T))
 
+    def test_write_hdf_str_column(self):
+        states = ct.SolutionArray(self.gas, 3, extra={'spam': 'eggs'})
+
+        outfile = pjoin(self.test_work_dir, 'solutionarray.h5')
+        states.write_hdf(outfile, mode='w')
+
+        b = ct.SolutionArray(self.gas, extra={'spam'})
+        b.read_hdf(outfile)
+        self.assertEqual(list(states.spam), list(b.spam))
+
+
 class TestRestoreIdealGas(utilities.CanteraTest):
     """ Test restoring of the IdealGas class """
     @classmethod

interfaces/cython/cantera/test/test_thermo.py

@@ -1787,6 +1787,16 @@ def test_purefluid(self):
         states.TP = np.linspace(400, 500, 5), 101325
         self.assertArrayNear(states.Q.squeeze(), np.ones(5))
 
+    def test_phase_of_matter(self):
+        water = ct.Water()
+        states = ct.SolutionArray(water, 5)
+        T = [300, 500, water.critical_temperature*2, 300]
+        P = [101325, 101325, 101325, water.critical_pressure*2]
+        states[:4].TP = T, P
+        states[4].TQ = 300, .4
+        pom = ['liquid', 'gas', 'supercritical', 'supercritical', 'liquid-gas-mix']
+        self.assertEqual(list(states.phase_of_matter), pom)
+
     def test_purefluid_getters(self):
         N = 11
         water = ct.Water()