Write tests for adapter pens

[adrientetar]

No description provided.

[anthrotype]

yea, it's a paradox the adapter pens are perhaps the most used classes from ufoLib but have no tests of their own. Hopefully they are tested "in the real world" :)
But yeah, we should add tests once we move ufoLib to fontTools.ufoLib.

[anthrotype]

just found a nasty bug in BasePointToSegmentPen..
for the special case of quadratic contours with no on-curve points, where it adds a special starting point with None coordinates, it also sets the kwargs to None.. But then if subclasses pass that kwargs to another pointpen's addPoint, using **dict unpacking, it will fail with TypeError as it's not a mapping.

[anthrotype]

relevant trufont issue here trufont/trufont#595

[anthrotype]

ufoLib/Lib/ufoLib/pointPen.py

Lines 119 to 123 in 49c7bd1

	if firstOnCurve is None:
	# Special case for quadratics: a contour with no on-curve
	# points. Add a "None" point. (See also the Pen protocol's
	# qCurveTo() method and fontTools.pens.basePen.py.)
	points.append((None, "qcurve", None, None, None))

[anthrotype]

just found a nasty bug in BasePointToSegmentPen..

well, it turns out it wasn't a bug, I just was using it improperly.
The cu2qu Cu2QuPointPen is a filter (point)pen that subclasses BasePointToSegmentPen; that's because the algorithm works on segments, not points; the super-class is used to collect points into lists of segments before converting the curves.
A point with None coordinates is a special marker that in the SegmentPen protocol means, this is a contour with no on-curve points, only quadratic off-curves (with implicit interpolated mid-points).
The Cu2QuPointPen should not forward this special None point on to the drawing pointpen, because a None point is not a valid point in the PointPen protocol. Points must always have (x, y) tuple.
Anyway mistery solved, ignore my previous posts.