make cell/sg optional. improve docstring

reciprocalspaceship/io/dials.py

@@ -84,7 +84,7 @@ def _concat(refl_data):
         ds = refl_data
     else:
         refl_data = [ds for ds in refl_data if ds is not None]
-        ds = rs.concat(refl_data)
+        ds = rs.concat(refl_data, check_isomorphous=False)
     expt_ids = set(ds.BATCH)
     LOGGER.debug(f"Found {len(ds)} refls from {len(expt_ids)} expts.")
     LOGGER.debug("Mapping batch column.")
@@ -191,30 +191,44 @@ def _read_dials_stills_ray(fnames, unitcell, spacegroup, numjobs=10, extra_cols=
 @spacegroupify
 def read_dials_stills(
     fnames,
-    unitcell,
-    spacegroup,
+    unitcell=None,
+    spacegroup=None,
     numjobs=10,
     parallel_backend=None,
     extra_cols=None,
     verbose=False,
     comm=None,
 ):
     """
+    Read reflections from still images processed by DIALS from fnames and return
+    them as a DataSet. This method does not convert columns to native rs MTZ dtypes.
+
     Parameters
     ----------
-    fnames: filenames
-    unitcell: unit cell tuple, Gemmi unit cell obj
-    spacegroup: space group symbol eg P4
-    numjobs: if backend==ray, specify the number of jobs (ignored if backend==mpi)
-    parallel_backend: ray, mpi, or None
-    extra_cols: list of additional column names to extract from the refltables. By default, this method will search for
+    fnames : list or tuple
+        A list or tuple of filenames (strings).
+    unitcell : gemmi.UnitCell or similar (optional)
+        The unit cell assigned to the returned dataset.
+    spacegroup : gemmi.SpaceGroup or similar (optional)
+        The spacegroup assigned to the returned dataset.
+    numjobs : int
+        If backend==ray, specify the number of jobs (ignored if backend==mpi). 
+    parallel_backend : string (optional)
+        "ray", "mpi", or None for serial. 
+    extra_cols : list (optional)
+        Optional list of additional column names to extract from the refltables. By default, this method will search for
         miller_index, id, s1, xyzcal.px, intensity.sum.value, intensity.sum.variance, delpsical.rad
-    verbose: whether to print stdout
-    comm: optionally override the communicator used by backend='mpi'
+    verbose : bool
+        Whether to print logging info to stdout
+    comm : mpi4py.MPI.Comm
+        Optionally override the communicator used by backend='mpi'
 
     Returns
     -------
-    rs dataset (pandas Dataframe)
+    ds : rs.DataSet 
+        The dataset containing reflection info aggregated from fnames. This method will not convert any of the
+        columns to native rs MTZ dtypes. DIALS data are natively double precision (64-bit). Converting to MTZ
+        will downcast them to 32-bit. Use ds.infer_mtz_dtypes() to convert to native rs dtypes if you required.
     """
     _set_logger(verbose)
 

tests/io/test_dials.py

@@ -106,7 +106,7 @@ def make_refls(unit_cell, sg, seed=8675309, file_prefix=""):
 def test_dials_reader(parallel_backend, verbose=False):
 
     unit_cell = 78, 78, 235, 90, 90, 120
-    sg = "P6522"
+    sg = "P 65 2 2"
     comm = None
     if parallel_backend == "mpi":
         comm = DummyComm()
@@ -162,6 +162,12 @@ def test_dials_reader(parallel_backend, verbose=False):
         assert np.allclose(df_m.I, df_m["intensity.sum.value"])
         assert np.allclose(df_m.varI, df_m["intensity.sum.sigma"] ** 2)
 
+        # Test that you don't need cell and symmetry to load the tables
+        ds =read_dials_stills(
+            pack_names, parallel_backend=None, numjobs=1, verbose=verbose
+        )
+        assert ds.cell is None
+        assert ds.spacegroup is None
 
 def test_verbosity():
     with tempfile.TemporaryDirectory() as tdir: