Use latest gala, add a catch for slicing half-loaded files

cogsworth/pop.py

@@ -216,6 +216,22 @@ def __getitem__(self, ind):
         # convert any Pandas Series to numpy arrays
         ind = ind.values if isinstance(ind, pd.Series) else ind
 
+        # if the population is associated with a file, make sure it's entirely loaded before slicing
+        if self._file is not None:
+            parts = ["initial_binaries", "bpp", "initial_galaxy", "orbits"]
+            vars = [self._initial_binaries, self._bpp, self._initial_galaxy, self._orbits]
+            masks = {f"has_{p}": False for p in parts}
+            with h5.File(self._file, "r") as f:
+                for p in parts:
+                    masks[f"has_{p}"] = p in f
+            missing_parts = [p for i, p in enumerate(parts) if (masks[f"has_{p}"] and vars[i] is None)]
+
+            if len(missing_parts) > 0:
+                raise ValueError(("This population was loaded from a file but you haven't loaded all parts "
+                                  "yet. You need to do this before indexing it. The missing parts are: "
+                                  f"{missing_parts}. You either need to access each of these variables or "
+                                  "reload the entire population using all parts."))
+
         # ensure indexing with the right type
         ALLOWED_TYPES = (int, slice, list, np.ndarray, tuple)
         if not isinstance(ind, ALLOWED_TYPES):
@@ -262,7 +278,6 @@ def __getitem__(self, ind):
                                  sampling_params=self.sampling_params,
                                  store_entire_orbits=self.store_entire_orbits)
         new_pop.n_binaries_match = new_pop.n_binaries
-        new_pop._file = self._file
 
         # proxy for checking whether sampling has been done
         if self._mass_binaries is not None:

cogsworth/tests/test_pop.py

@@ -3,6 +3,7 @@
 import cogsworth.pop as pop
 import cogsworth.sfh as sfh
 import cogsworth.observables as obs
+import h5py as h5
 import os
 import pytest
 
@@ -421,6 +422,24 @@ def test_indexing_mixed_types(self):
             it_worked = False
         self.assertFalse(it_worked)
 
+    def test_indexing_loaded_pop(self):
+        """Test indexing fails when trying to slice a half-loaded population"""
+        p = pop.Population(10)
+        p.perform_stellar_evolution()
+
+        with h5.File("DUMMY.h5", "w") as f:
+            f.create_dataset("orbits", data=[])
+        p._file = "DUMMY.h5"
+        p._orbits = None
+
+        it_worked = True
+        try:
+            p[:5]
+        except ValueError:
+            it_worked = False
+        self.assertFalse(it_worked)
+        os.remove("DUMMY.h5")
+
     def test_evolved_pop(self):
         """Check that the EvolvedPopulation class works as it should"""
         p = pop.Population(10)

setup.cfg

@@ -24,7 +24,7 @@ install_requires =
     astropy >= 5.0
     scipy >= 1.8
     pandas >= 2.1
-    gala @ git+https://github.com/adrn/gala
+    gala >= 1.9.1
     cosmic-popsynth >= 3.4.16
 
 [options.package_data]