Dataset: don't scale non-default units

qcfractal/interface/collections/dataset.py

@@ -11,11 +11,10 @@
 import numpy as np
 import pandas as pd
 import requests
-from tqdm import tqdm
-
 from pydantic import Field, validator
 from qcelemental import constants
 from qcelemental.models.types import Array
+from tqdm import tqdm
 
 from ..models import Citation, ComputeResponse, ObjectId, ProtoModel
 from ..statistics import wrap_statistics
@@ -39,12 +38,15 @@ class MoleculeEntry(ProtoModel):
 class ContributedValues(ProtoModel):
     name: str = Field(..., description="The name of the contributed values.")
     values: Any = Field(..., description="The values in the contributed values.")
-    index: Array[str] = Field(..., description="The entry index for the contributed values, matches the order of the `values` array.")
-    values_structure: Dict[str, Any] = Field({}, description="A machine readable description of the values structure. Typically not needed.")
+    index: Array[str] = Field(
+        ..., description="The entry index for the contributed values, matches the order of the `values` array.")
+    values_structure: Dict[str, Any] = Field(
+        {}, description="A machine readable description of the values structure. Typically not needed.")
 
     theory_level: Union[str, Dict[str, str]] = Field(..., description="A string representation of the theory level.")
     units: str = Field(..., description="The units of the values, can be any valid QCElemental unit.")
-    theory_level_details: Optional[Union[str, Dict[str, Optional[str]]]] = Field(None, description="A detailed reprsentation of the theory level.")
+    theory_level_details: Optional[Union[str, Dict[str, Optional[str]]]] = Field(
+        None, description="A detailed reprsentation of the theory level.")
 
     citations: Optional[List[Citation]] = Field(None, description="Citations associated with the contributed values.")
     external_url: Optional[str] = Field(None, description="An external URL to the raw contributed values data.")
@@ -59,6 +61,7 @@ def _make_array(cls, v):
 
         return v
 
+
 class Dataset(Collection):
     """
     The Dataset class for homogeneous computations on many molecules.
@@ -1171,8 +1174,23 @@ def units(self):
 
     @units.setter
     def units(self, value):
-
-        self.df *= constants.conversion_factor(self._units, value)
+        for column in self.df.columns:
+            try:
+                self.df[column] *= constants.conversion_factor(self._column_metadata[column]["units"], value)
+
+                # Cast units to quantities so that `kcal / mol` == `kilocalorie / mole`
+                metadata_quantity = constants.Quantity(self._column_metadata[column]["units"])
+                self_quantity = constants.Quantity(self._units)
+                if metadata_quantity != self_quantity:
+                    warnings.warn(f"Data column '{column}' did not have the same units as the dataset. "
+                                  f"This has been corrected.")
+                self._column_metadata[column]["units"] = value
+            except ValueError as e:
+                # This is meant to catch pint.errors.DimensionalityError without importing pint, which is too slow
+                if e.__class__.__name__ == "DimensionalityError":
+                    pass
+                else:
+                    raise
         self._units = value
 
     def set_default_program(self, program: str) -> bool:

qcfractal/tests/test_collections.py

@@ -439,6 +439,17 @@ def test_dataset_contributed_units(contributed_dataset_fixture):
     assert qcel.constants.ureg(
         ds._column_metadata[ds.get_values(name="Fake Dipole").columns[0]]["units"]) == qcel.constants.ureg("e * bohr")
 
+    old_units = ds.units
+    assert old_units == "kcal / mol"
+    ds.units = "kcal/mol"
+    before = ds.get_values()
+    ds.units = "hartree"
+    after = ds.get_values()
+    assert before["Fake Energy"][0] == after["Fake Energy"][0] / qcel.constants.conversion_factor(
+        "kcal/mol", "hartree")
+    assert before["Fake Gradient"][0][0, 0] == after["Fake Gradient"][0][0, 0]
+    ds.units = old_units
+
 
 def test_dataset_contributed_mixed_values(contributed_dataset_fixture):
     _, ds = contributed_dataset_fixture
@@ -531,6 +542,7 @@ def test_reactiondataset_check_state(fractal_compute_server):
 def test_contributed_dataset_values_subset(contributed_dataset_fixture, use_cache):
     client, ds = contributed_dataset_fixture
 
+    ds._clear_cache()
     allvals = ds.get_values()
     ds._clear_cache()
 
@@ -1061,6 +1073,7 @@ def test_s22_view_identical(s22_fixture):
     assert_view_identical(ds)
 
 
+@pytest.mark.slow
 def test_view_download_remote(s22_fixture):
     client, ds = s22_fixture
 
@@ -1118,11 +1131,13 @@ def test_contributed_dataset_plaintextview_write(contributed_dataset_fixture, tm
     ds.to_file(tmpdir / "test.tar.gz", "plaintext")
 
 
+@pytest.mark.slow
 def test_s22_dataset_plaintextview_write(s22_fixture, tmpdir):
     _, ds = s22_fixture
     ds.to_file(tmpdir / "test.tar.gz", "plaintext")
 
 
+@pytest.mark.slow
 def test_qm3_dataset_plaintextview_write(qm3_fixture, tmpdir):
     _, ds = qm3_fixture
     ds.to_file(tmpdir / "test.tar.gz", "plaintext")