Merge pull request #46 from KingsburyLab/serialize_fix

Fix Solution serialization

CHANGELOG.md

@@ -5,6 +5,14 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [Unreleased]
+
+### Fixed
+
+- Bugfix in `as_dict` to make serialization via `dumpfn` possible. Previously, `Quantity`
+  were not converted to a serializable form. Now, `Quantity` are converted to `str` in
+  `as_dict()`.
+
 ## [0.8.1] - 2023-10-01
 
 ### Changed

setup.cfg

@@ -53,7 +53,7 @@ install_requires =
     numpy
     scipy
     pint
-    pymatgen>2023.8.10
+    pymatgen>=2023.10.11
     iapws
     monty
     maggma

src/pyEQL/solution.py

@@ -1768,7 +1768,7 @@ def get_activity_coefficient(
         """
         Return the activity coefficient of a solute in solution.
 
-        The model used to calculte the activity coefficient is determined by the Solution's equation of state
+        The model used to calculate the activity coefficient is determined by the Solution's equation of state
         engine.
 
         Args:
@@ -2359,6 +2359,10 @@ def as_dict(self) -> dict:
         if self.volume_update_required:
             self._update_volume()
         d = super().as_dict()
+        for k, v in d.items():
+            # convert all Quantity to str
+            if isinstance(v, Quantity):
+                d[k] = str(v)
         # replace solutes with the current composition
         d["solutes"] = {k: f"{v} mol" for k, v in self.components.items()}
         # replace the engine with the associated str
@@ -2480,15 +2484,15 @@ def __add__(self, other: Solution):
     def __sub__(self, other: Solution):
         raise NotImplementedError("Subtraction of solutions is not implemented.")
 
-    def __mul__(self, factor: float | int):
+    def __mul__(self, factor: float):
         """
         Solution multiplication: scale all components by a factor. For example, Solution * 2 will double the moles of
         every component (including solvent). No other properties will change.
         """
         self.volume *= factor
         return self
 
-    def __truediv__(self, factor: float | int):
+    def __truediv__(self, factor: float):
         """
         Solution division: scale all components by a factor. For example, Solution / 2 will remove half of the moles
         of every compoonents (including solvent). No other properties will change.

src/pyEQL/utils.py

@@ -29,15 +29,7 @@ def standardize_formula(formula: str):
         charge number will always be listed explicitly and 2) the charge number will be enclosed in square brackets to remove any ambiguity in the meaning of the formula. For example, 'Na+', 'Na+1', and 'Na[+]' will all
         standardize to "Na[+1]"
     """
-    # TODO - hack to work around issues in pymatgen Ion.reduced_formula (until fixes can be merged upstream)
-    from pymatgen.util.string import charge_string
-
-    ion = Ion.from_formula(formula)
-    rform, factor = ion.get_reduced_formula_and_factor(hydrates=False)
-    charge = ion._charge / factor
-    chg_str = charge_string(charge)
-    return rform + chg_str
-    # return Ion.from_formula(formula).reduced_formula
+    return Ion.from_formula(formula).reduced_formula
 
 
 class FormulaDict(UserDict):

tests/test_solution.py

@@ -489,7 +489,7 @@ def test_arithmetic_and_copy(s2, s6):
         s2 + s_bad
 
 
-def test_serialization(s1, s2, s5):
+def test_as_from_dict(s1, s2):
     assert isinstance(s1.as_dict(), dict)
     s1_new = Solution.from_dict(s1.as_dict())
     assert s1_new.volume.magnitude == 2
@@ -530,3 +530,49 @@ def test_serialization(s1, s2, s5):
     # also should point to different Store instances
     # TODO currently this test will fail due to a bug in maggma's __eq__
     # assert s2_new.database != s2.database
+
+
+def test_serialization(s1, s2, tmpdir):
+    from monty.serialization import dumpfn, loadfn
+
+    dumpfn(s1, str(tmpdir / "s1.json"))
+    s1_new = loadfn(str(tmpdir / "s1.json"))
+    assert s1_new.volume.magnitude == 2
+    assert s1_new._solutes["H[+1]"] == "2e-07 mol"
+    assert s1_new.get_total_moles_solute() == s1.get_total_moles_solute()
+    assert s1_new.components == s1.components
+    assert np.isclose(s1_new.pH, s1.pH)
+    assert np.isclose(s1_new._pH, s1._pH)
+    assert np.isclose(s1_new.pE, s1.pE)
+    assert np.isclose(s1_new._pE, s1._pE)
+    assert s1_new.temperature == s1.temperature
+    assert s1_new.pressure == s1.pressure
+    assert s1_new.solvent == s1.solvent
+    assert s1_new._engine == s1._engine
+    # the solutions should point to different EOS instances
+    assert s1_new.engine != s1.engine
+    # also should point to different Store instances
+    # TODO currently this test will fail due to a bug in maggma's __eq__
+    # assert s1_new.database != s1.database
+
+    dumpfn(s2, str(tmpdir / "s2.json"))
+    s2_new = loadfn(str(tmpdir / "s2.json"))
+    assert s2_new.volume == s2.volume
+    # components concentrations should be the same
+    assert s2_new.components == s2.components
+    # but not point to the same instances
+    assert s2_new.components is not s2.components
+    assert s2_new.get_total_moles_solute() == s2.get_total_moles_solute()
+    assert np.isclose(s2_new.pH, s2.pH)
+    assert np.isclose(s2_new._pH, s2._pH)
+    assert np.isclose(s2_new.pE, s2.pE)
+    assert np.isclose(s2_new._pE, s2._pE)
+    assert s2_new.temperature == s2.temperature
+    assert s2_new.pressure == s2.pressure
+    assert s2_new.solvent == s2.solvent
+    assert s2_new._engine == s2._engine
+    # the solutions should point to different EOS instances
+    assert s2_new.engine != s2.engine
+    # also should point to different Store instances
+    # TODO currently this test will fail due to a bug in maggma's __eq__
+    # assert s2_new.database != s2.database