From 7d268fc74db727fd931619e5e5bc089e6fedf6fb Mon Sep 17 00:00:00 2001
From: Ingmar Schoegl <ischoegl@lsu.edu>
Date: Fri, 10 Jul 2020 14:46:32 -0500
Subject: [PATCH 1/5] [thermo] Enable import of previously exported str column
 from CSV files

Implementation based on a suggestion by Bryan W. Weber <bryan.w.weber@gmail.com>
---
 interfaces/cython/cantera/composite.py        | 27 +++++++++++++------
 .../cython/cantera/test/test_composite.py     | 10 +++++++
 2 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/interfaces/cython/cantera/composite.py b/interfaces/cython/cantera/composite.py
index f1172f9180a..6724fb3a2ed 100644
--- a/interfaces/cython/cantera/composite.py
+++ b/interfaces/cython/cantera/composite.py
@@ -4,6 +4,7 @@
 from ._cantera import *
 import numpy as np
 from collections import OrderedDict
+from distutils import version
 import csv as _csv
 
 import pkg_resources
@@ -955,14 +956,24 @@ 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 version.LooseVersion(np.__version__) < version.LooseVersion("1.14"):
+            # 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_:
+                    dtype = '{}'.format(data[label].dtype)
+                    dtype = dtype.replace('S', 'U').replace('|', '<')
+                    data_dict[label] = data[label].astype(dtype)
+                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):
diff --git a/interfaces/cython/cantera/test/test_composite.py b/interfaces/cython/cantera/test/test_composite.py
index fa75296afef..3c53f7c5586 100644
--- a/interfaces/cython/cantera/test/test_composite.py
+++ b/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.assertTrue((states.spam == b.spam).all)
+
     @utilities.unittest.skipIf(isinstance(_pandas, ImportError), "pandas is not installed")
     def test_to_pandas(self):
 

From d00a821ff81b32f6e9fa479fa99099d78c3408e1 Mon Sep 17 00:00:00 2001
From: Ingmar Schoegl <ischoegl@lsu.edu>
Date: Fri, 10 Jul 2020 16:06:52 -0500
Subject: [PATCH 2/5] [thermo] Enable export/import of str columns to/from HDF
 files

---
 interfaces/cython/cantera/composite.py        | 21 +++++++++++++------
 .../cython/cantera/test/test_composite.py     | 13 +++++++++++-
 2 files changed, 27 insertions(+), 7 deletions(-)

diff --git a/interfaces/cython/cantera/composite.py b/interfaces/cython/cantera/composite.py
index 6724fb3a2ed..a74f75f257c 100644
--- a/interfaces/cython/cantera/composite.py
+++ b/interfaces/cython/cantera/composite.py
@@ -963,8 +963,7 @@ def read_csv(self, filename):
             data_dict = OrderedDict()
             for label in data.dtype.names:
                 if data[label].dtype.type == np.bytes_:
-                    dtype = '{}'.format(data[label].dtype)
-                    dtype = dtype.replace('S', 'U').replace('|', '<')
+                    dtype = '<U' + data[label].dtype.__str__()[2:]
                     data_dict[label] = data[label].astype(dtype)
                 else:
                     data_dict[label] = data[label]
@@ -973,7 +972,7 @@ def read_csv(self, filename):
             data = np.genfromtxt(filename, delimiter=',',
                                  dtype=None, names=True, encoding=None)
             data_dict = OrderedDict({label: data[label]
-                                    for label in data.dtype.names})
+                                     for label in data.dtype.names})
         self.restore_data(data_dict)
 
     def to_pandas(self, cols=None, *args, **kwargs):
@@ -1126,8 +1125,13 @@ 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_:
+                    dtype = '|S' + value.dtype.__str__()[2:]
+                    dgroup.create_dataset(header, data=value.astype(dtype),
+                                          **hdf_kwargs)
+                else:
+                    dgroup.create_dataset(header, data=value, **hdf_kwargs)
 
         return group
 
@@ -1215,7 +1219,12 @@ def strip_ext(source):
             # load data
             data = OrderedDict()
             for name, value in dgroup.items():
-                if name != 'phase':
+                if name == 'phase':
+                    pass
+                elif value.dtype.type == np.bytes_:
+                    dtype = '<U' + value.dtype.__str__()[2:]
+                    data[name] = np.array(value).astype(dtype)
+                else:
                     data[name] = np.array(value)
 
         self.restore_data(data)
diff --git a/interfaces/cython/cantera/test/test_composite.py b/interfaces/cython/cantera/test/test_composite.py
index 3c53f7c5586..7c8346f2b3d 100644
--- a/interfaces/cython/cantera/test/test_composite.py
+++ b/interfaces/cython/cantera/test/test_composite.py
@@ -168,7 +168,7 @@ def test_write_csv_str_column(self):
 
         b = ct.SolutionArray(self.gas, extra={'spam'})
         b.read_csv(outfile)
-        self.assertTrue((states.spam == b.spam).all)
+        self.assertTrue((states.spam == b.spam).all())
 
     @utilities.unittest.skipIf(isinstance(_pandas, ImportError), "pandas is not installed")
     def test_to_pandas(self):
@@ -229,6 +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.assertTrue((states.spam == b.spam).all())
+
+
 class TestRestoreIdealGas(utilities.CanteraTest):
     """ Test restoring of the IdealGas class """
     @classmethod

From a0a19c1826d5796be2cc8ab2cf7919e90233eb92 Mon Sep 17 00:00:00 2001
From: Ingmar Schoegl <ischoegl@lsu.edu>
Date: Fri, 10 Jul 2020 16:39:16 -0500
Subject: [PATCH 3/5] [thermo] Enable phase_of_matter output for SolutionArray

---
 interfaces/cython/cantera/composite.py        |  7 +++++++
 interfaces/cython/cantera/test/test_thermo.py | 10 ++++++++++
 2 files changed, 17 insertions(+)

diff --git a/interfaces/cython/cantera/composite.py b/interfaces/cython/cantera/composite.py
index a74f75f257c..c6325d98d2a 100644
--- a/interfaces/cython/cantera/composite.py
+++ b/interfaces/cython/cantera/composite.py
@@ -448,6 +448,7 @@ class SolutionArray:
         # From Transport
         'viscosity', 'electrical_conductivity', 'thermal_conductivity',
     ]
+    _strings = ['phase_of_matter']
     _n_species = [
         # from ThermoPhase
         'Y', 'X', 'concentrations', 'partial_molar_enthalpies',
@@ -1326,6 +1327,9 @@ def _make_functions():
     def empty_scalar(self):
         return np.empty(self._shape)
 
+    def empty_strings(self):
+        return np.empty(self._shape, dtype='<U14')
+
     def empty_species(self):
         return np.empty(self._shape + (self._phase.n_selected_species,))
 
@@ -1352,6 +1356,9 @@ 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))
 
diff --git a/interfaces/cython/cantera/test/test_thermo.py b/interfaces/cython/cantera/test/test_thermo.py
index 7e67c84ad7f..987cab2c76e 100644
--- a/interfaces/cython/cantera/test/test_thermo.py
+++ b/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 = np.array(['liquid', 'gas', 'supercritical', 'supercritical', 'liquid-gas-mix'])
+        self.assertTrue((states.phase_of_matter == pom).all())
+
     def test_purefluid_getters(self):
         N = 11
         water = ct.Water()

From f1f83a11b3974301c57e2868b74334492a82ed26 Mon Sep 17 00:00:00 2001
From: Ingmar Schoegl <ischoegl@lsu.edu>
Date: Tue, 4 Aug 2020 10:08:46 -0500
Subject: [PATCH 4/5] [thermo] Streamline I/O support of string columns

Address review comments
---
 interfaces/cython/cantera/composite.py        | 13 ++++++-------
 interfaces/cython/cantera/test/test_thermo.py |  4 ++--
 2 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/interfaces/cython/cantera/composite.py b/interfaces/cython/cantera/composite.py
index c6325d98d2a..6cb0f2c1072 100644
--- a/interfaces/cython/cantera/composite.py
+++ b/interfaces/cython/cantera/composite.py
@@ -4,7 +4,6 @@
 from ._cantera import *
 import numpy as np
 from collections import OrderedDict
-from distutils import version
 import csv as _csv
 
 import pkg_resources
@@ -957,14 +956,14 @@ def read_csv(self, filename):
         using `restore_data`. This method allows for recreation of data
         previously exported by `write_csv`.
         """
-        if version.LooseVersion(np.__version__) < version.LooseVersion("1.14"):
+        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_:
-                    dtype = '<U' + data[label].dtype.__str__()[2:]
+                    dtype = '<U' + str(data[label].dtype)[2:]
                     data_dict[label] = data[label].astype(dtype)
                 else:
                     data_dict[label] = data[label]
@@ -1128,7 +1127,7 @@ def write_hdf(self, filename, *args, cols=None, group=None, subgroup=None,
                 dgroup.attrs[key] = val
             for header, value in data.items():
                 if value.dtype.type == np.str_:
-                    dtype = '|S' + value.dtype.__str__()[2:]
+                    dtype = '|S' + str(value.dtype)[2:]
                     dgroup.create_dataset(header, data=value.astype(dtype),
                                           **hdf_kwargs)
                 else:
@@ -1221,9 +1220,9 @@ def strip_ext(source):
             data = OrderedDict()
             for name, value in dgroup.items():
                 if name == 'phase':
-                    pass
+                    continue
                 elif value.dtype.type == np.bytes_:
-                    dtype = '<U' + value.dtype.__str__()[2:]
+                    dtype = '<U' + str(value.dtype)[2:]
                     data[name] = np.array(value).astype(dtype)
                 else:
                     data[name] = np.array(value)
@@ -1328,7 +1327,7 @@ def empty_scalar(self):
         return np.empty(self._shape)
 
     def empty_strings(self):
-        return np.empty(self._shape, dtype='<U14')
+        return np.empty(self._shape, dtype='<U50')
 
     def empty_species(self):
         return np.empty(self._shape + (self._phase.n_selected_species,))
diff --git a/interfaces/cython/cantera/test/test_thermo.py b/interfaces/cython/cantera/test/test_thermo.py
index 987cab2c76e..0cea69dc691 100644
--- a/interfaces/cython/cantera/test/test_thermo.py
+++ b/interfaces/cython/cantera/test/test_thermo.py
@@ -1794,8 +1794,8 @@ def test_phase_of_matter(self):
         P = [101325, 101325, 101325, water.critical_pressure*2]
         states[:4].TP = T, P
         states[4].TQ = 300, .4
-        pom = np.array(['liquid', 'gas', 'supercritical', 'supercritical', 'liquid-gas-mix'])
-        self.assertTrue((states.phase_of_matter == pom).all())
+        pom = ['liquid', 'gas', 'supercritical', 'supercritical', 'liquid-gas-mix']
+        self.assertEqual(list(states.phase_of_matter), pom)
 
     def test_purefluid_getters(self):
         N = 11

From e4e7f3d04e861a2c1b0931269d50009961acc9d4 Mon Sep 17 00:00:00 2001
From: Ingmar Schoegl <ischoegl@lsu.edu>
Date: Thu, 6 Aug 2020 18:06:14 -0500
Subject: [PATCH 5/5] [thermo] Add comment for maximum length of strings in
 SolutionArray

The length of strings assigned by built-in SolutionArray methods is
limited to 50 characters.
---
 interfaces/cython/cantera/composite.py           | 14 +++++++-------
 interfaces/cython/cantera/test/test_composite.py |  4 ++--
 2 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/interfaces/cython/cantera/composite.py b/interfaces/cython/cantera/composite.py
index 6cb0f2c1072..b614bf67cc9 100644
--- a/interfaces/cython/cantera/composite.py
+++ b/interfaces/cython/cantera/composite.py
@@ -963,8 +963,7 @@ def read_csv(self, filename):
             data_dict = OrderedDict()
             for label in data.dtype.names:
                 if data[label].dtype.type == np.bytes_:
-                    dtype = '<U' + str(data[label].dtype)[2:]
-                    data_dict[label] = data[label].astype(dtype)
+                    data_dict[label] = data[label].astype('U')
                 else:
                     data_dict[label] = data[label]
         else:
@@ -1127,8 +1126,7 @@ def write_hdf(self, filename, *args, cols=None, group=None, subgroup=None,
                 dgroup.attrs[key] = val
             for header, value in data.items():
                 if value.dtype.type == np.str_:
-                    dtype = '|S' + str(value.dtype)[2:]
-                    dgroup.create_dataset(header, data=value.astype(dtype),
+                    dgroup.create_dataset(header, data=value.astype('S'),
                                           **hdf_kwargs)
                 else:
                     dgroup.create_dataset(header, data=value, **hdf_kwargs)
@@ -1222,8 +1220,7 @@ def strip_ext(source):
                 if name == 'phase':
                     continue
                 elif value.dtype.type == np.bytes_:
-                    dtype = '<U' + str(value.dtype)[2:]
-                    data[name] = np.array(value).astype(dtype)
+                    data[name] = np.array(value).astype('U')
                 else:
                     data[name] = np.array(value)
 
@@ -1327,7 +1324,10 @@ def empty_scalar(self):
         return np.empty(self._shape)
 
     def empty_strings(self):
-        return np.empty(self._shape, dtype='<U50')
+        # 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,))
diff --git a/interfaces/cython/cantera/test/test_composite.py b/interfaces/cython/cantera/test/test_composite.py
index 7c8346f2b3d..974d822aa09 100644
--- a/interfaces/cython/cantera/test/test_composite.py
+++ b/interfaces/cython/cantera/test/test_composite.py
@@ -168,7 +168,7 @@ def test_write_csv_str_column(self):
 
         b = ct.SolutionArray(self.gas, extra={'spam'})
         b.read_csv(outfile)
-        self.assertTrue((states.spam == b.spam).all())
+        self.assertEqual(list(states.spam), list(b.spam))
 
     @utilities.unittest.skipIf(isinstance(_pandas, ImportError), "pandas is not installed")
     def test_to_pandas(self):
@@ -237,7 +237,7 @@ def test_write_hdf_str_column(self):
 
         b = ct.SolutionArray(self.gas, extra={'spam'})
         b.read_hdf(outfile)
-        self.assertTrue((states.spam == b.spam).all())
+        self.assertEqual(list(states.spam), list(b.spam))
 
 
 class TestRestoreIdealGas(utilities.CanteraTest):