Avoid division by 0 in Plot Profile.

[bpindelski]

This PR fixes the situation in which a ZeroDivisionError is thrown when the ROI line is parallel to the X axis and hence the y1 and y2 values are equal, returning 0 in the subtraction.

The error returned was:

Traceback (most recent call last):
  File "./script", line 453, in <module>
    fileAnns, message = processImages(conn, scriptParams)
  File "./script", line 382, in processImages
    processLines(conn, scriptParams, image, lines, lineWidth, f)
  File "./script", line 264, in processLines
    lineWidth, theZ, theC, theT)
  File "./script", line 93, in getLineData
    rads = math.atan(float(lineX) / lineY)
ZeroDivisionError: float division

To test - check that the values returned from the script make sense. /cc @will-moore

[joshmoore]

Hmmm.... I have a feeling this should ring some Python version bell, but perhaps we've already moved passed that minimum requirement. (I know it doesn't work with 2.4)

[bpindelski]

@joshmoore You are right, this is a Python 2.5 feature. And I still remember the lesson that using post-2.4 features taught me. Having said that, please see http://www.openmicroscopy.org/site/support/omero5/sysadmins/system-requirements.html.

[sbesson]

My immediate thought is is there a reason why are we sticking to math.atan(y/x) instead of math.atan2(y,x) which is correctly handling zero values for x and or y (and would prevent our code from handling this error?)

[sbesson]

Apart from my comment above, the fix proposed works when running Plot_Profile on horizontal single lines of various widths.

[joshmoore]

Next steps?

[bpindelski]

@joshmoore I'll leave the call to @will-moore. I've tried looking at the code of the script, but it doesn't immediately say if X/Y-parallel lines are allowed.

[will-moore]

The Plot Profile script now works fine on horizontal and vertical lines. The fix above should also be applied to the Kymograph script to fix the same bug there, see: http://trac.openmicroscopy.org.uk/ome/ticket/11988 Thanks!

[bpindelski]

@will-moore Thanks for testing. I'm happy to go with @sbesson's suggestion in this PR and use atan2, as long as it won't impact the values produced by the script.

[joshmoore]

Last call? Otherwise, I'll merge and we can atan2 later.

[will-moore]

Fine by me ;)

[bpindelski]

--rebased-to #90