get_fastccd_images() does not work with roi given as input

[tankonst]

When roi coordinates are passed to
get_fastccd_images(h, (bgnd8, bgnd2, bgnd1), flat=ff, tag='fccd_image', roi = (0, 0, 50, 50))
the function returns an AttributeError.

AttributeError Traceback (most recent call last)
in

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/csxtools/utils.py in get_fastccd_images(light_header, dark_headers, flat, gain, tag, roi)
92
93 tt = ttime.time()
---> 94 b = get_images_to_3D(bgnd_events, dtype=np.uint16)
95 logger.info("Image conversion took %.3f seconds",
96 ttime.time() - tt)

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/csxtools/utils.py in get_images_to_3D(images, dtype)
173
174 """
--> 175 im = np.vstack([np.asarray(im, dtype=dtype) for im in images])
176 return im
177

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/csxtools/utils.py in (.0)
173
174 """
--> 175 im = np.vstack([np.asarray(im, dtype=dtype) for im in images])
176 return im
177

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/slicerator/init.py in (.0)
472
473 def iter(self):
--> 474 return (self._get(i) for i in range(len(self)))
475
476 def getitem(self, i):

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/slicerator/init.py in _get(self, key)
459 # We need to copy here: else any _proc_func that acts inplace would
460 # change the ancestor value.
--> 461 return self._proc_func(*(copy(a[key]) for a in self._ancestors))
462
463 def repr(self):

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/slicerator/init.py in proc_func(*x)
683 if all_pipe:
684 def proc_func(x):
--> 685 return func((x + args), **kwargs)
686
687 return Pipeline(proc_func, *ancestors)

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/csxtools/utils.py in _crop_images(image, roi)
207 @pipeline
208 def _crop_images(image, roi):
--> 209 return _crop(image, roi)
210
211

/opt/conda_envs/analysis-2019-3.0-ml/lib/python3.7/site-packages/csxtools/utils.py in _crop(image, roi)
214 # Assuming ROI is specified in the "rotated" (correct) orientation
215 roi = [image_shape[-2]-roi[3], roi[0], image_shape[-1]-roi[1], roi[2]]
--> 216 return image.T[roi[1]:roi[3], roi[0]:roi[2]].T
217
218

AttributeError: 'ImageStack' object has no attribute 'T'

[ambarb]

@tankonst thanks for making it clear that this is an issue. see #68 for more discussion. I think we agreed first with DAMA that we would wait until the data handler was fixed for data broker catalogs worked. It does now. The next step is to implement catalogs instead of get_images.

It is probably worth planning on how were are going to do this and also roll in metadata from image concatenation (not associated with the roi functionality in this issue, but with the fccd settings to remove overscan and adjust for framestore versus non-framestore).

I think it is worth doing these first so that we only have to deal with this issue here one more time, instead of two times. Does that make sense, or do you need this functionality for the issue cited here now?

@danielballan @leofang I tag you because I know you also have interests at play.

[tankonst]

@ambarb yes, it makes sense. I was just thinking that it probably should be faster for for users to load only part of the image (aka 'area1' in the current notebook) for analysis of different scans in the same experiment. But this is not a top priority. I am closing it then since the issue was discussed in #68 (sorry, I haven't noticed before posting).

[ambarb]

@tankonst I don't mind leaving it open. It makes it clear to anyone else new to the repo. The discussion is reference, but hard to trek through.