unit tests

[LiangJYu]

This PR adds unit tests to the 3 modules in s1-reader. The tests are currently very sparse and serve as a frame to build on.

Usage:

1. checkout unit_test branch from my fork
2. cd to tests directory
3. in terminal call: pytest

To do:

• test and validate more attributes

[yunjunz]

Thank you @LiangJYu, following your note above, the test runs nicely.

How did you prepare the SAFE zip file? I notice there are measurement folder and tiff files but with 0 in size. Maybe write this in a README.md file within the tests folder?

[LiangJYu]

Thank you @LiangJYu, following your note above, the test runs nicely.

How did you prepare the SAFE zip file? I notice there are measurement folder and tiff files but with 0 in size. Maybe write this in a README.md file within the tests folder?

The 0 size files were touchd with a correct name and path. I will add a README to tests folder.

[LiangJYu]

Thus far, the checks for simple attributes have been implemented. I'm unsure how to test the following more complex attributes.

orbit: Ensure all state vectors within sensing start and stop? Check if position actually above surface of the earth? Check if velocity fast enough?

poly1D: These vary with burst so the verbose method would be to store and check order, std, mean, and coeffs of each. I can't think of a cleaner way.

lut2D: Can this be skiped since this is derived from poly1d?

[hfattahi]

@LiangJYu One test that I can think of for orbit is the following where we compute the coordinates for the middle of the burst and it should fall within the burst border.

burst = bursts[0]

from shapely.geometry import Point

# middle of the burst in radar coordinates
r0 = burst.starting_range + 0.5*burst.width*burst.range_pixel_spacing
t0 = (isce3.core.DateTime(burst.sensing_mid)-burst.orbit.reference_epoch)

dem = isce3.geometry.DEMInterpolator(0)
dop = 0.0
rdr_grid=burst.as_isce3_radargrid()
llh = isce3.geometry.rdr2geo(t0.total_seconds(), r0, burst.orbit, rdr_grid.lookside, dop, rdr_grid.wavelength, dem)
pnt = Point(np.degrees(llh[0]), np.degrees(llh[1]))
assert(burst.border[0].contains(pnt))

Even with the inaccurate orbit within the annotation file, we should be able to pass this test.

[hfattahi]

@LiangJYu for the lut2d you can check if we are able to eval successfully compared to poly1d. For example :

np.isclose(burst.doppler.lut2d.eval(t0.total_seconds(),r0), burst.doppler.poly1d.eval(r0))

t0 and r0 can be the same middle of swath in my previous comment.

[LiangJYu]

@hfattahi @yunjunz @vbrancat With the new tests added, I think this PR is ready for another look. Thanks!

[hfattahi]

LGTM! Thanks @LiangJYu for addressing my comments.

[yunjunz]

Looks great! Thank you @LiangJYu.

[seongsujeong]

Hi @LiangJYu @yunjunz I am about to merge the annotation reader in PR #48. The new implementation loads IPF version from manifest.safe file in SAFE .zip file, which is currently not included in the test data. It would be great if the test data is updated with manifest.safe file included, so that the new PR can pass CircieCI.