lycantropos / hypothesis_geometry Goto Github PK

I'm proposing to add a strategy for linestrings, since it's quite a common geometry object used in GIS and other geometry-related applications. See, for example, LineString in SQL docs and LineString in Shapely docs.

Also see the wikipedia entry on linestrings.

We need a strategy for contours with vertices visible from centroid.

This might exist but I don't see it in the documentation.

It would be amazing if strategies that generate closed results were able to compute their own perimeters and areas. I'm writing a great deal of bounding box handling code at the moment, and i'd love more rigorous testing of how we handle that.

We need a strategy for triplets of disjoint multipoints, multisegments and multipolygons.

We should have strategy for sequences of non-overlapping polygons.

As it is currently written convex_contours uses convex hull of points which doesn't scale well for cases with sizes limits.

We can probably use an algorithm described here.

Currently we have naive O(n^2) algorithm, some kind of O(n log n) sweep-line-algorithm could be used instead.

We should have a strategy for polygons constructed as list of non-intersecting contours (with some of them marked as holes).

We need a strategy for sequences of distinctive points.

Since axis-aligned rectangles are very common in practice we need a separate strategy for them.

We should have a strategy for sequences of non-crossing & non-overlapping segments.