roman-snpit-snappl 0.2.3__tar.gz → 0.3.0__tar.gz

{roman_snpit_snappl-0.2.3 → roman_snpit_snappl-0.3.0}/.github/workflows/run_snappl_tests.yml

@@ -42,7 +42,7 @@ jobs:
 
       - name: pull latest SNPIT cpu image
         run: |
-          docker pull rknop/roman-snpit-env:cpu
+          docker pull rknop/roman-snpit-env:cpu-dev
 
       - name: run test
         run: |

{roman_snpit_snappl-0.2.3 → roman_snpit_snappl-0.3.0}/MANIFEST.in

@@ -8,4 +8,5 @@ global-exclude *.pyc *.o
 prune snappl/_version.py
 include LICENSE
 include README.rst
-include pyproject.toml
+include pyproject.toml
+prune snappl/tests

{roman_snpit_snappl-0.2.3/roman_snpit_snappl.egg-info → roman_snpit_snappl-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: roman_snpit_snappl
-Version: 0.2.3
+Version: 0.3.0
 Summary: Photometry utilities for the Roman SNPIT
 Author: Roman Supernove Project Infrastructure Team
 Maintainer-email: Roman SN PIT <raknop@lbl.gov>

roman_snpit_snappl-0.3.0/changes/14.snappl.rst

@@ -0,0 +1 @@
+Remove tests from what gets packaged for distribution

roman_snpit_snappl-0.3.0/changes/15.feature.rst

@@ -0,0 +1 @@
+Write tests for wcs.py and image.py, and in so doing fix a bunch of problems.  Bit of image.py refactoring.

roman_snpit_snappl-0.3.0/changes/16.feature.rst

@@ -0,0 +1 @@
+Add support for galsim wCSes

{roman_snpit_snappl-0.2.3 → roman_snpit_snappl-0.3.0/roman_snpit_snappl.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: roman_snpit_snappl
-Version: 0.2.3
+Version: 0.3.0
 Summary: Photometry utilities for the Roman SNPIT
 Author: Roman Supernove Project Infrastructure Team
 Maintainer-email: Roman SN PIT <raknop@lbl.gov>

{roman_snpit_snappl-0.2.3 → roman_snpit_snappl-0.3.0}/roman_snpit_snappl.egg-info/SOURCES.txt

@@ -26,6 +26,9 @@ tox.ini
 changes/.gitkeep
 changes/10.snappl.rst
 changes/13.bugfix.rst
+changes/14.snappl.rst
+changes/15.feature.rst
+changes/16.feature.rst
 changes/3.snappl.rst
 changes/5.snappl.rst
 changes/8.snappl.rst
@@ -55,14 +58,4 @@ snappl/psf.py
 snappl/wcs.py
 snappl/_dev/__init__.py
 snappl/_dev/scm_version.py
-snappl/data/README.rst
-snappl/tests/__init__.py
-snappl/tests/conftest.py
-snappl/tests/docker-compose.yaml
-snappl/tests/test_opensim2024fitsimage.py
-snappl/tests/test_oversampled_image_psf.py
-snappl/tests/test_version_string.py
-snappl/tests/test_yaml_serialized_oversampled_image_psf.py
-snappl/tests/image_test_data/Roman_TDS_simple_model_F184_662_11.fits.gz
-snappl/tests/image_test_data/test_cutout.npy
-snappl/tests/psf_test_data/testpsfarray.npz
+snappl/data/README.rst

{roman_snpit_snappl-0.2.3 → roman_snpit_snappl-0.3.0}/snappl/_version.py

@@ -17,5 +17,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '0.2.3'
-__version_tuple__ = version_tuple = (0, 2, 3)
+__version__ = version = '0.3.0'
+__version_tuple__ = version_tuple = (0, 3, 0)

roman_snpit_snappl-0.3.0/snappl/image.py

@@ -0,0 +1,506 @@
+import types
+
+import numpy as np
+from astropy.io import fits
+from astropy.nddata.utils import Cutout2D
+# from astropy.coordinates import SkyCoord
+
+from snpit_utils.logger import SNLogger
+from snappl.wcs import AstropyWCS, GalsimWCS
+
+
+class Exposure:
+    pass
+
+
+class OpenUniverse2024Exposure:
+    def __init__( self, pointing ):
+        self.pointing = pointing
+
+
+# ======================================================================
+# The base class for all images.  This is not useful by itself, you need
+#   to instantiate a subclass.  However, everything that you call on an
+#   object you instantiate should have its interface defined in this
+#   class.
+
+class Image:
+    """Encapsulates a single 2d image."""
+
+    data_array_list = [ 'all', 'data', 'noise', 'flags' ]
+
+    def __init__( self, path, exposure, sca ):
+        """Instantiate an image.  You probably don't want to do that.
+
+        This is an abstract base class that has limited functionality.
+        You probably want to instantiate a subclass.
+
+        For all implementations, the properties data, noise, and flags
+        are lazy-loaded.  That is, they start empty, but when you access
+        them, an internal buffer gets loaded with that data.  This means
+        it can be very easy for lots of memory to get used without your
+        realizing it.  There are a couple of solutions.  The first, is
+        to call Image.free() when you're sure you don't need the data
+        any more, or if you know you want to get rid of it for a while
+        and re-read it from disk later.  The second is just not to
+        access the data, noise, and flags properties, instead use
+        Image.get_data(), and manage the data object lifetime yourself.
+
+        Parameters
+        ----------
+          path : str
+            Path to image file, or otherwise some kind of indentifier
+            that allows the class to find the image.
+
+          exposure : Exposure (or instance of Exposure subclass)
+            The exposure this image is associated with, or None if it's
+            not associated with an Exposure (or youdon't care)
+
+          sca : int
+            The Sensor Chip Assembly that would be called the
+            chip number for any other telescope but is called SCA for
+            Roman.
+
+        """
+        self.inputs = types.SimpleNamespace()
+        self.inputs.path = path
+        self.inputs.exposure = exposure
+        self.inputs.sca = sca
+        self._wcs = None      # a BaseWCS object (in wcs.py)
+        self._is_cutout = False
+
+    @property
+    def data( self ):
+        """The image data, a 2d numpy array."""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement data" )
+
+    @data.setter
+    def data( self, new_value ):
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement data setter" )
+
+    @property
+    def noise( self ):
+        """The 1σ pixel noise, a 2d numpy array."""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement noise" )
+
+    @noise.setter
+    def noise( self, new_value ):
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement noise setter" )
+
+    @property
+    def flags( self ):
+        """An integer 2d numpy array of pixel masks / flags TBD"""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement flags" )
+
+    @flags.setter
+    def flags( self, new_value ):
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement flags setter" )
+
+    @property
+    def image_shape( self ):
+        """Tuple: (ny, nx) pixel size of image."""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement image_shape" )
+
+    @property
+    def sky_level( self ):
+        """Estimate of the sky level in ADU."""
+        raise NotImplementedError( "Do.")
+
+    @property
+    def exptime( self ):
+        """Exposure time in seconds."""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement exptime" )
+
+    @property
+    def band( self ):
+        """Band (str)"""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement band" )
+
+    @property
+    def mjd( self ):
+        """MJD of the start of the image (defined how? TAI?)"""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement mjd" )
+
+    @property
+    def position_angle( self ):
+        """Position angle in degrees east of north (or what)?"""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement position_angle" )
+
+    def fraction_masked( self ):
+        """Fraction of pixels that are masked."""
+        raise NotImplementedError( "Do.")
+
+    def get_data( self, which='all', always_reload=False, cache=False ):
+        """Read the data from disk and return one or more 2d numpy arrays of data.
+
+        Parameters
+        ----------
+          which : str
+            What to read:
+              'data' : just the image data
+              'noise' : just the noise data
+              'flags' : just the flags data
+              'all' : data, noise, and flags
+
+          always_reload: bool, default False
+            Whether this is supported depends on the subclass.  If this
+            is false, then get_data() has the option of returning the
+            values of self.data, self.noise, and/or self.flags instead
+            of always loading the data.  If this is True, then
+            get_data() will ignore the self._data et al. properties.
+
+          cache: bool, default False
+            Normally, get_data() just reads the data and does not do any
+            internal caching.  If this is True, and the subclass
+            supports it, then the object will cache the loaded data so
+            that future calls with always_reload will not need to reread
+            the data, nor will accessing the data, noise, and flags
+            properties.
+
+        The data read not stored in the class, so when the caller goes
+        out of scope, the data will be freed (unless the caller saved it
+        somewhere.  This does mean it's read from disk every time.
+
+        Returns
+        -------
+          list (length 1 or 3 ) of 2d numpy arrays
+
+        """
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement get_data" )
+
+
+    def free( self ):
+        """Try to free memory."""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement free" )
+
+    def get_wcs( self, wcsclass=None ):
+        """Get image WCS.  Will be an object of type BaseWCS (from wcs.py) (really likely a subclass).
+
+        Parameters
+        ----------
+          wcsclass : str or None
+            By default, the subclass of BaseWCS you get back will be
+            defined by the Image subclass of the object you call this
+            on.  If you want a specific subclass of BaseWCS, you can put
+            the name of that class here.  It may not always work; not
+            all types of images are able to return all types of wcses.
+
+        Returns
+        -------
+          object of a subclass of snappl.wcs.BaseWCS
+
+        """
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement get_wcs" )
+
+    def get_cutout(self, ra, dec, size):
+        """Make a cutout of the image at the given RA and DEC.
+
+        Returns
+        -------
+          snappl.image.Image
+        """
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement get_cutout" )
+
+
+    @property
+    def coord_center(self):
+        """[RA, DEC] (both floats) in degrees at the center of the image"""
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement coord_center" )
+
+
+# ======================================================================
+# Lots of classes will probably internally store all of data, noise, and
+#   flags as 2d numpy arrays.  Common code for those classes is here.
+
+class Numpy2DImage( Image ):
+    def __init__( self, *args, **kwargs ):
+        super().__init__( *args, **kwargs )
+
+        self._data = None
+        self._noise = None
+        self._flags = None
+        self._image_shape = None
+
+    @property
+    def data( self ):
+        if self._data is None:
+            self._load_data()
+        return self._data
+
+    @data.setter
+    def data(self, new_value):
+        if ( isinstance(new_value, np.ndarray)
+             and np.issubdtype(new_value.dtype, np.floating)
+             and len(new_value.shape) ==2
+            ):
+            self._data = new_value
+        else:
+            raise TypeError( "Data must be a 2d numpy array of floats." )
+
+    @property
+    def noise( self ):
+        if self._noise is None:
+            self._load_data()
+        return self._noise
+
+    @noise.setter
+    def noise( self, new_value ):
+        if ( isinstance( new_value, np.ndarray )
+             and np.issubdtype( new_value.dtype, np.floating )
+             and len( new_value.shape ) == 2
+            ):
+            self._noise = new_value
+        else:
+            raise TypeError( "Noise must be a 2d numpy array of floats." )
+
+    @property
+    def flags( self ):
+        if self._flags is None:
+            self._load_data()
+        return self._flags
+
+    @flags.setter
+    def flags( self, new_value ):
+        if ( isinstance( new_value, np.ndarray )
+             and np.issubdtype( new_value.dtype, np.integer )
+             and len( new_value.shape ) == 2
+            ):
+            self._flags = new_value
+        else:
+            raise TypeError( "Flags must be a 2d numpy array of integers." )
+
+    @property
+    def image_shape( self ):
+        """Subclasses probably want to override this!
+
+        This implementation accesses the .data property, which will load the data
+        from disk if it hasn't been already.  Actual images are likely to have
+        that information availble in a manner that doesn't require loading all
+        the image data (e.g. in a header), so subclasses should do that.
+
+        """
+        if self._image_shape is None:
+            self._image_shape = self.data.shape
+        return self._image_shape
+
+    def _load_data( self ):
+        """Loads (or reloads) the data from disk."""
+        imgs = self.get_data()
+        self._data = imgs[0]
+        self._noise = imgs[1]
+        self._flags = imgs[2]
+
+    def free( self ):
+        self._data = None
+        self._noise = None
+        self._flags = None
+
+
+# ======================================================================
+# A base class for FITSImages which use an AstropyWCS wcs.  Not useful
+#   by itself, because which image you load will have different
+#   assumptions about which HDU holds image, weight, flags, plus header
+#   information will be different etc.  However, there will be some
+#   shared code between all FITS implementations, so that's here.
+
+class FITSImage( Numpy2DImage ):
+    def __init__( self, *args, **kwargs ):
+        super().__init__( *args, **kwargs )
+
+        self._data = None
+        self._noise = None
+        self._flags = None
+        self._header = None
+
+    @property
+    def image_shape(self):
+        """tuple: (ny, nx) shape of image"""
+
+        if not self._is_cutout:
+            hdr = self._get_header()
+            self._image_shape = ( hdr['NAXIS1'], hdr['NAXIS2'] )
+            return self._image_shape
+
+        if self._image_shape is None:
+            self._image_shape = self.data.shape
+
+        return self._image_shape
+
+    @property
+    def coord_center(self):
+        """[ RA and Dec ] at the center of the image."""
+
+        wcs = self.get_wcs()
+        return wcs.pixel_to_world( self.image_shape[1] //2, self.image_shape[0] //2 )
+
+    def _get_header( self ):
+        raise NotImplementedError( f"{self.__class__.__name__} needs to implement _get_header()" )
+
+    def get_wcs( self, wcsclass=None ):
+        wcsclass = "AstropyWCS" if wcsclass is None else wcsclass
+        if ( self._wcs is None ) or ( self._wcs.__class__.__name__ != wcsclass ):
+            if wcsclass == "AstropyWCS":
+                hdr = self._get_header()
+                self._wcs = AstropyWCS.from_header( hdr )
+            elif wcsclass == "GalsimWCS":
+                hdr = self._get_header()
+                self._wcs = GalsimWCS.from_header( hdr )
+        return self._wcs
+
+    def get_cutout(self, x, y, xsize, ysize=None):
+        """Creates a new snappl image object that is a cutout of the original image, at a location in pixel-space.
+
+        This implementation (in FITSImage) assumes that the image WCS is an AstropyWCS.
+
+        Parameters
+        ----------
+        x : int
+            x pixel coordinate of the center of the cutout.
+        y : int
+            y pixel coordinate of the center of the cutout.
+        xsize : int
+            Width of the cutout in pixels.
+        ysize : int
+            Height of the cutout in pixels. If None, set to xsize.
+
+        Returns
+        -------
+        cutout : snappl.image.Image
+            A new snappl image object that is a cutout of the original image.
+
+        """
+        if not all( [ isinstance( x, (int, np.integer) ),
+                      isinstance( y, (int, np.integer) ),
+                      isinstance( xsize, (int, np.integer) ),
+                      ( ysize is None or isinstance( ysize, (int, np.integer) ) )
+                     ] ):
+            raise TypeError( "All of x, y, xsize, and ysize must be integers." )
+
+        if ysize is None:
+            ysize = xsize
+        if xsize % 2 != 1 or ysize % 2 != 1:
+            raise ValueError( f"Size must be odd for a well defined central "
+                              f"pixel, you tried to pass a size of {xsize, ysize}.")
+
+        SNLogger.debug(f'Cutting out at {x , y}')
+        data, noise, flags = self.get_data( 'all', always_reload=False )
+
+        wcs = self.get_wcs()
+        if ( wcs is not None ) and ( not isinstance( wcs, AstropyWCS ) ):
+            raise TypeError( "Error, FITSImage.get_cutout only works with AstropyWCS wcses" )
+        apwcs = None if wcs is None else wcs._wcs
+
+        # Remember that numpy arrays are indexed [y, x] (at least if they're read with astropy.io.fits)
+        astropy_cutout = Cutout2D(data, (x, y), size=(ysize, xsize), mode='strict', wcs=apwcs)
+        astropy_noise = Cutout2D(noise, (x, y), size=(ysize, xsize), mode='strict', wcs=apwcs)
+        astropy_flags = Cutout2D(flags, (x, y), size=(ysize, xsize), mode='strict', wcs=apwcs)
+
+        snappl_cutout = self.__class__(self.inputs.path, self.inputs.exposure, self.inputs.sca)
+        snappl_cutout._data = astropy_cutout.data
+        snappl_cutout._wcs = None if wcs is None else AstropyWCS( astropy_cutout.wcs )
+        snappl_cutout._noise = astropy_noise.data
+        snappl_cutout._flags = astropy_flags.data
+        snappl_cutout._is_cutout = True
+
+        return snappl_cutout
+
+    def get_ra_dec_cutout(self, ra, dec, xsize, ysize=None):
+        """Creates a new snappl image object that is a cutout of the original image, at a location in pixel-space.
+
+        Parameters
+        ----------
+        ra : float
+            RA coordinate of the center of the cutout, in degrees.
+        dec : float
+            DEC coordinate of the center of the cutout, in degrees.
+        xsize : int
+            Width of the cutout in pixels.
+        ysize : int
+            Height of the cutout in pixels. If None, set to xsize.
+
+        Returns
+        -------
+        cutout : snappl.image.Image
+            A new snappl image object that is a cutout of the original image.
+        """
+
+        wcs = self.get_wcs()
+        x, y = wcs.world_to_pixel( ra, dec )
+        x = int( np.floor( x + 0.5 ) )
+        y = int( np.floor( y + 0.5 ) )
+        return self.get_cutout( x, y, xsize, ysize )
+
+
+# ======================================================================
+# OpenUniverse 2024 Images are gzipped FITS files
+#  HDU 0 : (something, no data)
+#  HDU 1 : SCI (32-bit float)
+#  HDU 2 : ERR (32-bit float)
+#  HDU 3 : DQ (32-bit integer)
+
+class OpenUniverse2024FITSImage( FITSImage ):
+    def __init__( self, *args, **kwargs ):
+        super().__init__( *args, **kwargs )
+
+    def get_data( self, which='all', always_reload=False, cache=False ):
+        if self._is_cutout:
+            raise RuntimeError( "get_data called on a cutout image, this will return the ORIGINAL UNCUT image. "
+                                "Currently not supported.")
+        if which not in Image.data_array_list:
+            raise ValueError( f"Unknown which {which}, must be all, data, noise, or flags" )
+
+        if not always_reload:
+            if ( ( which == 'all' )
+                 and ( self._data is not None )
+                 and ( self._noise is not None )
+                 and ( self._flags is not None )
+                ):
+                return [ self._data, self._noise, self._flags ]
+
+            if ( which == 'data' ) and ( self._data is not None ):
+                return [ self._data ]
+
+            if ( which == 'noise' ) and ( self._noise is not None ):
+                return [ self._noise ]
+
+            if ( which == 'flags' ) and ( self._flags is not None ):
+                return [ self._flags ]
+
+        SNLogger.info( f"Reading FITS file {self.inputs.path}" )
+        with fits.open( self.inputs.path ) as hdul:
+            if cache:
+                self._header = hdul[1].header
+            if which == 'all':
+                imgs = [ hdul[1].data, hdul[2].data, hdul[3].data ]
+                if cache:
+                    self._data = imgs[0]
+                    self._noise = imgs[1]
+                    self._flags = imgs[2]
+                return imgs
+            elif which == 'data':
+                if cache:
+                    self._data = hdul[1].data
+                return [ hdul[1].data ]
+            elif which == 'noise':
+                if cache:
+                    self._noise = hdul[2].data
+                return [ hdul[2].data ]
+            elif which == 'flags':
+                if cache:
+                    self._flags = hdul[3].data
+                return [ hdul[3].data ]
+            else:
+                raise RuntimeError( f"{self.__class__.__name__} doesn't understand data plane {which}" )
+
+    def _get_header(self):
+        """Get the header of the image."""
+        if self._header is None:
+            with fits.open(self.inputs.path) as hdul:
+                self._header = hdul[1].header
+        return self._header
+
+    @property
+    def band(self):
+        """The band the image is taken in (str)."""
+        header = self._get_header()
+        return header['FILTER'].strip()