Lazy import pandas in Structure.as_dataframe() to improve startup speed (#3568)

* lazy import pandas in Structure.as_dataframe() to improve startup speed

* fix allclose removal in scipy 1.12

Author: janosh

pymatgen/core/structure.py

@@ -24,7 +24,6 @@
 from typing import TYPE_CHECKING, Any, Callable, Literal, SupportsIndex, cast, get_args
 
 import numpy as np
-import pandas as pd
 from monty.dev import deprecated
 from monty.io import zopen
 from monty.json import MSONable
@@ -2625,16 +2624,19 @@ def as_dataframe(self):
             0    (Si)  0.0  0.0  0.000000e+00  0.0  0.000000e+00  0.000000e+00       5
             1    (Si)  0.0  0.0  1.000000e-7  0.0 -2.217138e-7  3.135509e-7      -5
         """
-        data = []
+        # pandas lazy imported for speed (https://github.com/materialsproject/pymatgen/issues/3563)
+        import pandas as pd
+
+        data: list[list[str | float]] = []
         site_properties = self.site_properties
         prop_keys = list(site_properties)
         for site in self:
             row = [site.species, *site.frac_coords, *site.coords]
-            for k in prop_keys:
-                row.append(site.properties.get(k))
+            for key in prop_keys:
+                row.append(site.properties.get(key))
             data.append(row)
 
-        df = pd.DataFrame(data, columns=["Species", "a", "b", "c", "x", "y", "z", *prop_keys])
+        df = pd.DataFrame(data, columns=["Species", *"abcxyz", *prop_keys])
         df.attrs["Reduced Formula"] = self.composition.reduced_formula
         df.attrs["Lattice"] = self.lattice
         return df

tests/core/test_structure.py

@@ -105,9 +105,12 @@ def test_get_orderings(self):
         assert sqs[0].formula == "Mn8 Fe8"
 
     def test_as_dataframe(self):
-        df = self.propertied_structure.as_dataframe()
-        assert df.attrs["Reduced Formula"] == self.propertied_structure.composition.reduced_formula
-        assert df.shape == (2, 8)
+        df_struct = self.propertied_structure.as_dataframe()
+        assert df_struct.attrs["Reduced Formula"] == self.propertied_structure.composition.reduced_formula
+        assert df_struct.shape == (2, 8)
+        assert list(df_struct) == ["Species", *"abcxyz", "magmom"]
+        assert list(df_struct["magmom"]) == [5, -5]
+        assert list(map(str, df_struct["Species"])) == ["Si1", "Si1"]
 
     def test_equal(self):
         struct = self.struct

tests/electronic_structure/test_plotter.py

@@ -7,7 +7,6 @@
 
 import matplotlib.pyplot as plt
 import numpy as np
-import scipy
 from matplotlib import rc
 from numpy.testing import assert_allclose
 from pytest import approx
@@ -269,13 +268,15 @@ def test_bz_plot(self):
         )
 
     def test_fold_point(self):
-        assert scipy.allclose(
+        assert_allclose(
             fold_point([0.0, -0.5, 0.5], lattice=self.rec_latt),
             self.rec_latt.get_cartesian_coords([0.0, 0.5, 0.5]),
+            atol=1e-8,
         )
-        assert scipy.allclose(
+        assert_allclose(
             fold_point([0.1, -0.6, 0.2], lattice=self.rec_latt),
             self.rec_latt.get_cartesian_coords([0.1, 0.4, 0.2]),
+            atol=1e-8,
         )