Add customizable invalid geometry handling #46
Conversation
| @@ -56,13 +56,28 @@ def transform(shape, func): | |||
| raise ValueError('Unknown geometry type, "%s"' % shape.type) | |||
|
|
|||
|
|
|||
| def on_invalid_geometry_raise(shape): | |||
| raise Exception('Invalid geometry: %s' % shape.wkt) | |||
There was a problem hiding this comment.
Do we want to raise plain Exception, or a more specific class such as ValueError or RuntimeError (or even a class of our own)?
There was a problem hiding this comment.
Good point. I like ValueError. Updated in 9660d52.
There are cases where the validation handler can generate points that when rounded will also generate invalid geometry. We repeat the entire rounding, winding, and validation process to catch these scenarios.
|
@zerebubuth the test failure is due to a rounding behavior change in python 3, specifically that it rounds to the nearest even number for ties instead of always up. This sounds like more balanced rounding behavior to me; do you think that we would want the same behavior with python 2 as well? |
| for x in valid_geometries] | ||
| self.failUnless(shape1.is_valid) | ||
| self.failUnless(shape2.is_valid) | ||
| self.failUnless(shape1.area > 0) |
There was a problem hiding this comment.
FailUnless is deprecated. Here you want to use assertGreater which will also be easier to debug in case the test fails.
|
Thanks @loicgasser |
|
Whoops, hit enter too early. Updated in 3e7e4f2. |
|
@rmarianski I'm not sure if the rounding behaviour fairness is going to matter much in terms of the tile coordinates, but it's definitely going to be less confusing to be consistent across Python versions! |
|
@zerebubuth do you think this can be merged in now? Or are there still other scenarios to consider? |
|
👍 |
Connects to #45
@zerebubuth could you review please?