CASA Image loader

doc/faq.rst

@@ -59,6 +59,24 @@ set the ``QT_API`` environment variable to either ``pyqt`` or ``pyside``, depend
     QT_API=pyside notebook --gui qt
 
 
+Does Glue Understand CASA Cubes?
+--------------------------------
+Many radio astronomy datacubes are 4 dimensional, and give intensity
+as a function of 2 spatial dimensions, a frequency dimension, and a
+Stokes polarization dimension. By default, Glue will read these images
+in as 4D hypercubes.
+
+However, you might wish to load the image as a series of 3D cubes,
+one for each Stokes parameter. This would more easily allow you,
+for example, to compare histograms of intensities for each Stokes
+parameter.
+
+Because of this, Glue has a special "CASA image loader", to load
+these kinds of files. To open a file in this mode, select "CASA PPV Cube"
+in the file type dropdown when opening a file. This will create a single
+Data object with 1 attribute for each Stokes parameter.
+
+
 Something is broken, or confusing. What should I do?
 ----------------------------------------------------
 If you think you've found a bug in Glue, feel free to add

glue/core/data_factories.py

@@ -241,6 +241,48 @@ def is_gridded_data(filename, **kwargs):
 set_default_factory('hdf5', gridded_data)
 
 
+def casalike_cube(filename, **kwargs):
+    """
+    This provides special support for 4D CASA-like cubes,
+    which have 2 spatial axes, a spectral axis, and a stokes axis
+    in that order.
+
+    Each stokes cube is split out as a separate component
+    """
+    result = Data()
+    with fits.open(filename, **kwargs) as hdulist:
+        array = hdulist[0].data
+        header = hdulist[0].header
+    result.coords = coordinates_from_header(header)
+    for i in range(array.shape[0]):
+        result.add_component(array[[i]], label='STOKES %i' % i)
+    return result
+
+
+def is_casalike(filename, **kwargs):
+    """
+    Check if a file is a CASA like cube,
+    with (P,P,V,Stokes) layout
+    """
+    if not is_fits(filename):
+        return False
+    with fits.open(filename) as hdulist:
+        if len(hdulist) != 1:
+            return False
+        if hdulist[0].header['NAXIS'] != 4:
+            return False
+
+        from astropy.wcs import WCS
+        w = WCS(hdulist[0].header)
+
+    ax = [a.get('coordinate_type') for a in w.get_axis_types()]
+    return ax == ['celestial', 'celestial', 'spectral', 'stokes']
+
+
+casalike_cube.label = 'CASA PPV Cube'
+casalike_cube.identifier = is_casalike
+
+
 def _ascii_identifier_v02(origin, args, kwargs):
     # this works for astropy v0.2
     if isinstance(args[0], basestring):
@@ -419,3 +461,4 @@ def img_data(file_name):
 
 __factories__.append(img_data)
 __all__.append('img_data')
+__factories__.append(casalike_cube)

glue/core/tests/test_data_factories.py

@@ -140,3 +140,15 @@ def test_column_spaces():
         d = df.load_data(fname)
     assert d['a'].dtype == np.float
     np.testing.assert_array_equal(d['a'], [np.nan, 2])
+
+
+def test_casalike():
+    data = 'x\xda\xed\x98Qo\xa3F\x10\xc7\xfbQ\xe6\xcd\x89\xce`v\xd9] R\x1f0\xde$$\xd8\xf8\x80Xq_*.!\xa9%\x1b\xa7@zM?}w\xc1v.\x97\xf8\xae\xc6\xfb\xd0\x07\xfe\xf2\x83e\xc9?\r\xf3\x1fvf6\xf6\xc7\xd3\x80\x03\xfc\n\x1f(\x81\x01\xdc\xad\xf3\x87u\xb1*\xa1Z\xc3\xb9\x9f\xc4PVi~\x9f\x16\xf7\xf0\xa1\x86~2\xf5o?\xe4i\x8c\x08^Z\x14\xe9\x0b\xdc\xa7U\n\xd5\xcbS\x06?\xd1\xc4\xbd\xf5c\xd8\x13\x9f\xe4\xe5\xcf\xab/Y\x01\xeb\x87-y\xb1\xca\xf2r\xb1\xce\xcb\xfd<\xb4\x87\x87\xe1`\xd5<\xac\x98g*\xe6\x91\xbd\xf9;X\xc3\xb1{\x05\x1b?\x98\xce,\xc2lF\x0cj:\x84kF+\x9e?y\xe5\x99\x14[\x84\xda\x98\x98\x8e\xd1\x927ua\xc3\xd3\x88Nm\xdbf\xb6i"\xc7p\xf8\'\x03\xb5\xe0%\xf3)\xafy=?\xafdeU/=h\xafpx\xc5\xbd\xa4\xe1y\x00&\xc1p\x0cN\xb9\x867\x13?i\x9e\xf7j>\x18rw\xfc&\xbe\xc1\xb0X<\xfeQ\xe5YY\xc2\xc9\xd3\xe2\xefly\n\xcf\xf9\xa2\xda\xcb\xe3\x9fo\xfcIx[\xfb\x8bu\xe3\x1b\t?\xcc\xc3\xe3\x8b\xdc\x11\x8f\xe7\xb1\x8c\xef\xfc\x9a\xd6?\x1d\x95\xbf \x9cLCq\x00\xca\xf8\x90n\xbf\x8d\xaf\xc5\xfb\x16\xb8\xc9\x8e\xc7\x04\xcfa\xc8\xa0\x94:\x94\xb5\xab\xbf\xa9g\x18H|\x9a\xf8\xbe\xcb\x9f\xd1\x8e\x877<C\x11\xcfT\xcc#\x8ay"\x7fX%\x0foxH]\xfe\xb0\xe2\xfca\xc5\xf93\x15\xe7\xcfT\\\x7f\xa6J?\x88\xe2\xf8D\xfe\x88\xe2\xfc\x11\xc5\xf9#\x8a\xeb\x8f\xa8\xf2\xc3\x93\xfd\xb7\x9e\xd7z\x91\xabiZ\xecO\x8e:\xef\xbdh\xe6\x06\x9b\xf9\x8f\x8a\xf9\xc5\xc0\x88 f!j\xb4;\x9f\xbd\x11\x0f\x92\x86\xa71\xdd!\xaf\x12\xf3\x0bm\x13\x9f\x18\x9fw\xf1!\xf3\xd8~\xe4\xc9~\xde\xe4\xef>{<\xbe_\xd6~\xd4\xf3no\xc4=E~`P\xd6/k?v<E~`\xc5~`\xc5~\xd4\xfbBo\x16N\x93\xe3y\xd2\x8ff\xff\xd0\xf0n\x1e\xc2\x04!\xf1\xbc\xb4\xa5\x1f\x9b}\xa69\x0f\x10c\x16\xc3\xd4\x12<\xd2\xd2\x8f7\xbc#\xcf\x17\xe9G\x93\xbf\xd5\xa0T\xe4G\xbdo\xf5\xe2$\xbc\xe6\xb1\n?\x88\xc2\xe7\x95~\xa8\xe4I?\x88b?\x9a\xfcm\x7f:*\x7f\xd3\x19\xfe\x1dm\xf6K%\xfdM\xf0\xb0B^\xc4\xe3\xe4<\xfa\\\xf3L\x9dP\x87\x99\xf6\x96\x87\x0c\x18DYY\xc1y\x91\xfd\xf9\x9c\xe5w/pr\xf9\xcf\xe9\x0fy\xf1\x94{\x9b\xfdh\xe8Fs\x8fO\x927\xfb[\xfc\x94\xddUE\xba\x84"{\xc8\n\xc1\xcc\xe0\xa1HW{\xefa\xdc \x91\x15\xb8\x8d\x8fa\xdb\xc2\x98"j\xa3&>wYeE\x9eV\x12\xb3\r\xf2\x15\xfdW\xba|\xce\xde\xf1\xe4\x05\xd1\x9ez\xf9\t\xaf\xde8\xdf\xf0f<\x88\xf8\xf9\xde\xfb\x92\x01\x82 \x8e\xfa\xe2\xdb%\x0f\xfa`\x8a\xfd;\xee\xc3\'L\x19D\xe9\xfdb\xfd\xbe\xfe\xc2\xf1X\xe4\x0c\xee\xd22\xbd[\x17\x19\xe4\xeb\\\xdb]y=\x97\xe9cv\x06D@G}\xa0p\xc1\xc3>0\x88\xc3\x9b>Xp!\xf2\xf4\xee\n\x8d\x07<\xf6\xc2\xa9\xf0\x83\xcf\x02\xf7\xe8z\x16\xf1\xf3h\xc6#\xd9\x7f\x0b\x1d\xdc\x95\x8c\x0c\xdc\xdd\xb7\xeb\xeci\x99\x1dpC1r\x13\xae\t\xa8\xe0a\x03!M|\xb0\x95\x18\xe8\x8c\x18g\x86\xa5\xd3\xda\x9a\x03"N\xfc\xf1v?\xbfI<P\xf1\xbc\x91\x0b\xea\xe65\xc1\x13s\x8b\xbayC\xf0D\x11h\xb7\xb2_"\x9d\x89\x14R\xd1\xdfL\xf1\x9a\x08\x1ek\xcd\x9bK\x1e\xd5\r\x82\x1c\xdbvlf0v\x14\xef\xb7\xe6\xfd\xa5\x94\xd8\x96\x83Mfc\xd4\x8e\'\xeb\xa5\xb9_\x93\xf5\x82\xc5L%J&A\xd6\x191\xcf(\xd6\x19\xa5u\xbd\x0cF\xf2\x15\xae\xef\x8b\x1f\x16\xcb\x0c\xbe\xa6%|-\x16U\x95\xe5\xdf\xc5\x17\xf9\x17\xf2\x06P\xf0<7v\xe5\x19\xa3\x1bp"\xde\xb8/\xe2o\x05r\x18\xc3\xa7\x07T\xd0\xa5/\xdan4\x87\x1a\x16\'n\x94@\x10^$\xee0\xe0\xad\xea\x8fOF\xd0\xa9S\xa7N\x9d:u\xea\xd4\xa9\xd3\xffE\xbft\xea\xd4\xa9S\xa7N\x9d:u\xea\xd4\xe9?\xeb_\xdc?$\x07'
+    with make_file(data, '.fits', decompress=True) as fname:
+        d = df.load_data(fname, factory=df.casalike_cube)
+
+    assert d.shape == (1, 2, 2, 2)
+    d['STOKES 0']
+    d['STOKES 1']
+    d['STOKES 2']
+    d['STOKES 3']