index error get_untiled_crop with a tifffile input

[ViriatoII]

I get an index error when using a normal tifffile input.

IndexError                                Traceback (most recent call last)
---> 26 _get_untiled_crop(page, 300, 200, 10, 10)

/tmp/ipykernel_460/304543409.py in _get_untiled_crop(page, i0, j0, h, w)
     12 
     13         for index in range(i0, i1):
---> 14             offset = page.dataoffsets[index]
     15             bytecount = page.databytecounts[index]
     16 

IndexError: tuple index out of range

The index generating error is a 300, my desired i0 coordinate. The .dataoffsets instance does not correspond to what you expect in an input page

My page input is simply read from a tifffile handle. It seems to work

>page = tifffile.TiffFile(test_slide).pages[0]
>page.imagewidth, page.imagelength
(2142, 2134)

My page.dataoffsets only has 34 elements, it goes like this:
(8, 274184, 548360, ....)

So what's causing this difference and how can I solve it?

Cheers,
Ricardo

[HowcanoeWang]

Dear Ricardo,

Thanks for using our package, we have understood that you meet the open tifffile problem?

Do you mind if you would like to share the code that meets the error 26 _get_untiled_crop(page, 300, 200, 10, 10)?

[ViriatoII]

Hey HawcanoeWang,

That was fast, thanks!

def _get_untiled_crop(page, i0, j0, h, w):

        im_width = page.imagewidth
        im_height = page.imagelength

        i1, j1 = i0 + h, j0 + w
        if i0 < 0 or j0 < 0 or i1 >= im_height or j1 >= im_width:
            raise ValueError(f"Requested crop area is out of image bounds.{i0}_{i1}_{im_height}, {j0}_{j1}_{im_width}")
        
        out = np.empty((page.imagedepth, h, w, page.samplesperpixel), dtype=page.dtype)
        fh = page.parent.filehandle

        for index in range(i0, i1):
            offset = page.dataoffsets[index]
            bytecount = page.databytecounts[index]

            fh.seek(offset)
            data = fh.read(bytecount)

            tile, indices, shape = page.decode(data, index)

            out[:,index-i0,:,:] = tile[:,:,j0:j1,:]

        return out[0,:,:,:]
    
test_slide =  "33042_W0A1_P1X_R2_RO-Channel1_27052022-21-33-59_s1.ome.tiff"
page = tifffile.TiffFile(test_slide).pages[0]
_get_untiled_crop(page, 300, 200, 10, 10) 

Would you like the test_slide file as well?
It's a 280Mb tiff file. Reading it with tifffile.imread generates a correct array with shape (32, 2134, 2142) (32 is the z layers). I noticed tifffile.TiffFile() generated 32 pages, so I assume each page is a z layer.

[HowcanoeWang]

could you please check page.is_tiled? If it is tiled, i guess you need to use _get_tiled_crop instead.

BTW, it is not recommended to call hidden functions start with '_' directly. Please use easyidp.geotiff.tifffile_crop(page, 300, 200, 10, 10) instead. (please use the latest code by git clone, rather than pip install easyidp)

[HowcanoeWang]

could you please check page.is_tiled? If it is tiled, i guess you need to use _get_tiled_crop instead.

BTW, it is not recommended to call hidden functions start with '_' directly. Please use easyidp.geotiff.tifffile_crop(page, 300, 200, 10, 10) instead. (please use the latest code by git clone, rather than pip install easyidp)

your _get_untiled_crop is code from v1.0, we have fixed some bug in it and please update to v2.0

[ViriatoII]

Oh, yes, _get_untiled_crop is defined twice in geotiff.py. I was copying the first one but the second one works!

(and the page was not tiled).

Thank you very much for the attention!
Cheers,
Ricardo

[HowcanoeWang]

Glad that works for you!
PS: _get_untiled_crop is not defined twice! They are _get_tiled_crop and _get_untiled_crop, dealing with page.is_tiled==True and page.is_tiled==False respectively.