Allow access of the filenames from the dataset object

changelog/56.feature.1.rst

@@ -0,0 +1,2 @@
+Allow easy access to the filenames contained in an
+`dkist.io.BaseFITSArrayContainer` object via a `~dkist.io.BaseFITSArrayContainer.filenames` property.

changelog/56.feature.2.rst

@@ -0,0 +1 @@
+`dkist.io.BaseFITSArrayContainer` objects are now sliceable.

dkist/dataset/dataset.py

@@ -81,7 +81,7 @@ def __init__(self, data, uncertainty=None, mask=None, wcs=None,
         else:
             self.missing_axis = missing_axis
 
-        self.array_container = None
+        self._array_container = None
 
     @classmethod
     def from_directory(cls, directory):
@@ -128,9 +128,16 @@ def from_asdf(cls, filepath):
             raise TypeError(f"This file is not a valid DKIST asdf file, it fails validation with: {e.message}.")
 
         cls = cls(data, wcs=wcs)
-        cls.array_container = array_container
+        cls._array_container = array_container
         return cls
 
+    @property
+    def array_container(self):
+        """
+        A reference to the files containing the data.
+        """
+        return self._array_container
+
     @property
     def pixel_axes_names(self):
         if self.wcs.input_frame:

dkist/dataset/tests/test_dataset.py

@@ -118,3 +118,12 @@ def test_no_wcs_slice(dataset):
 def test_crop_few_slices(dataset_4d):
     sds = dataset_4d[0, 0]
     assert len(sds.wcs.input_frame.axes_order)
+
+
+def test_array_container():
+    dataset = Dataset.from_directory(os.path.join(rootdir, 'EIT'))
+    assert dataset.array_container is dataset._array_container
+    with pytest.raises(AttributeError):
+        dataset.array_container = 10
+
+    assert len(dataset.array_container.filenames) == 11

dkist/io/fits.py

@@ -51,14 +51,16 @@ def __repr__(self):
         if self._array is None:
             return "<FITS array (unloaded) in {0} shape: {1} dtype: {2}>".format(
                 self.fitsarray.fileuri, self.fitsarray.shape, self.fitsarray.dtype)
-        return repr(self._array)
+        return "<FITS array (loaded) in {0} shape: {1} dtype: {2}>\n{3!r}".format(
+            self.fitsarray.fileuri, self.fitsarray.shape, self.fitsarray.dtype, self._array)
 
     def __str__(self):
         # str alone should not force loading of the data
         if self._array is None:
             return "<FITS array (unloaded) in {0} shape: {1} dtype: {2}>".format(
                 self.fitsarray.fileuri, self.fitsarray.shape, self.fitsarray.dtype)
-        return str(self._array)
+        return "<FITS array (loaded) in {0} shape: {1} dtype: {2}>\n{3!r}".format(
+            self.fitsarray.fileuri, self.fitsarray.shape, self.fitsarray.dtype, self._array)
 
     def __array__(self):
         return self.fits_array

dkist/io/reference_collections.py

@@ -37,9 +37,7 @@ class BaseFITSArrayContainer(metaclass=abc.ABCMeta):
     """
 
     def __init__(self, reference_array, *, loader, **kwargs):
-
         reference_array = np.asarray(reference_array, dtype=object)
-
         self._check_contents(reference_array)
 
         # If the first dimension is one we are going to squash it.
@@ -52,7 +50,10 @@ def __init__(self, reference_array, *, loader, **kwargs):
         loader_array = np.empty_like(reference_array, dtype=object)
         for i, ele in enumerate(reference_array.flat):
             loader_array.flat[i] = loader(ele, **kwargs)
+
         self.loader_array = loader_array
+        self._loader = partial(loader, **kwargs)
+        self.reference_array = reference_array
 
     def _check_contents(self, reference_array):
         """
@@ -67,6 +68,19 @@ def _check_contents(self, reference_array):
             assert ele.dtype == dtype
             assert ele.shape == shape
 
+    def __getitem__(self, item):
+        return type(self)(self.reference_array[item], loader=self._loader)
+
+    @property
+    def filenames(self):
+        """
+        Return a list of file names referenced by this Array Container.
+        """
+        names = []
+        for ear in self.reference_array.flat:
+            names.append(ear.fileuri)
+        return names
+
     @abc.abstractproperty
     def array(self):
         """

dkist/io/tests/test_collections.py

@@ -24,6 +24,30 @@ def externalarray():
         return f.tree['data']
 
 
+def test_slicing(externalarray):
+    ac = NumpyFITSArrayContainer(externalarray, loader=AstropyFITSLoader, basepath=eitdir)
+    ext_shape = np.array(externalarray, dtype=object).shape
+    assert ac.loader_array.shape == ext_shape
+    assert ac.shape == tuple(list(ext_shape) + [128, 128])
+
+    assert isinstance(ac.array, np.ndarray)
+    assert_allclose(ac.array, np.array(ac))
+
+    ac = ac[5:8]
+    ext_shape = np.array(externalarray[5:8], dtype=object).shape
+    assert ac.loader_array.shape == ext_shape
+    assert ac.shape == tuple(list(ext_shape) + [128, 128])
+
+    assert isinstance(ac.array, np.ndarray)
+    assert_allclose(ac.array, np.array(ac))
+
+
+def test_filenames(externalarray):
+    ac = NumpyFITSArrayContainer(externalarray, loader=AstropyFITSLoader, basepath=eitdir)
+    assert len(ac.filenames) == len(externalarray)
+    assert ac.filenames == [e.fileuri for e in externalarray]
+
+
 def test_numpy(externalarray):
     ac = NumpyFITSArrayContainer(externalarray, loader=AstropyFITSLoader, basepath=eitdir)
     ext_shape = np.array(externalarray, dtype=object).shape

dkist/io/tests/test_fits.py

@@ -59,8 +59,8 @@ def test_array(absolute_fl):
     assert isinstance(absolute_fl.fits_header, fits.Header)
 
     for contain in ("efz20040301.000010_s.fits", str(absolute_fl.shape), absolute_fl.dtype):
-        assert contain not in repr(absolute_fl)
-        assert contain not in str(absolute_fl)
+        assert contain in repr(absolute_fl)
+        assert contain in str(absolute_fl)
 
 
 def test_nan():