Merge pull request #2846 from fonttools/issue2843
[varLib.models] Attempt to fix #2843 by computing the axis ranges
This commit is contained in:
Just van Rossum
GitHub
commit
d102b7a9fb
@ -115,12 +115,15 @@ def normalizeLocation(location, axes):
|
||||
return out
|
||||
|
||||
|
||||
def supportScalar(location, support, ot=True, extrapolate=False):
|
||||
def supportScalar(location, support, ot=True, extrapolate=False, axisRanges=None):
|
||||
"""Returns the scalar multiplier at location, for a master
|
||||
with support. If ot is True, then a peak value of zero
|
||||
for support of an axis means "axis does not participate". That
|
||||
is how OpenType Variation Font technology works.
|
||||
|
||||
If extrapolate is True, axisRanges must be a dict that maps axis
|
||||
names to (axisMin, axisMax) tuples.
|
||||
|
||||
>>> supportScalar({}, {})
|
||||
1.0
|
||||
>>> supportScalar({'wght':.2}, {})
|
||||
@ -137,11 +140,17 @@ def supportScalar(location, support, ot=True, extrapolate=False):
|
||||
0.75
|
||||
>>> supportScalar({'wght':2.5, 'wdth':.5}, {'wght':(0,2,4), 'wdth':(-1,0,+1)})
|
||||
0.75
|
||||
>>> supportScalar({'wght':4}, {'wght':(0,2,3)}, extrapolate=True)
|
||||
2.0
|
||||
>>> supportScalar({'wght':4}, {'wght':(0,2,2)}, extrapolate=True)
|
||||
2.0
|
||||
>>> supportScalar({'wght':3}, {'wght':(0,1,2)}, extrapolate=True, axisRanges={'wght':(0, 2)})
|
||||
-1.0
|
||||
>>> supportScalar({'wght':-1}, {'wght':(0,1,2)}, extrapolate=True, axisRanges={'wght':(0, 2)})
|
||||
-1.0
|
||||
>>> supportScalar({'wght':3}, {'wght':(0,2,2)}, extrapolate=True, axisRanges={'wght':(0, 2)})
|
||||
1.5
|
||||
>>> supportScalar({'wght':-1}, {'wght':(0,2,2)}, extrapolate=True, axisRanges={'wght':(0, 2)})
|
||||
-0.5
|
||||
"""
|
||||
if extrapolate and axisRanges is None:
|
||||
raise TypeError("axisRanges must be passed when extrapolate is True")
|
||||
scalar = 1.0
|
||||
for axis, (lower, peak, upper) in support.items():
|
||||
if ot:
|
||||
@ -160,18 +169,19 @@ def supportScalar(location, support, ot=True, extrapolate=False):
|
||||
continue
|
||||
|
||||
if extrapolate:
|
||||
if v < -1 and lower <= -1:
|
||||
if peak <= -1 and peak < upper:
|
||||
axisMin, axisMax = axisRanges[axis]
|
||||
if v < axisMin and lower <= axisMin:
|
||||
if peak <= axisMin and peak < upper:
|
||||
scalar *= (v - upper) / (peak - upper)
|
||||
continue
|
||||
elif -1 < peak:
|
||||
elif axisMin < peak:
|
||||
scalar *= (v - lower) / (peak - lower)
|
||||
continue
|
||||
elif +1 < v and +1 <= upper:
|
||||
if +1 <= peak and lower < peak:
|
||||
elif axisMax < v and axisMax <= upper:
|
||||
if axisMax <= peak and lower < peak:
|
||||
scalar *= (v - lower) / (peak - lower)
|
||||
continue
|
||||
elif peak < +1:
|
||||
elif peak < axisMax:
|
||||
scalar *= (v - upper) / (peak - upper)
|
||||
continue
|
||||
|
||||
@ -189,9 +199,8 @@ def supportScalar(location, support, ot=True, extrapolate=False):
|
||||
class VariationModel(object):
|
||||
"""Locations must have the base master at the origin (ie. 0).
|
||||
|
||||
If the extrapolate argument is set to True, then location values are
|
||||
interpretted in the normalized space, ie. in the [-1,+1] range, and
|
||||
values are extrapolated outside this range.
|
||||
If the extrapolate argument is set to True, then values are extrapolated
|
||||
outside the axis range.
|
||||
|
||||
>>> from pprint import pprint
|
||||
>>> locations = [ \
|
||||
@ -241,6 +250,7 @@ class VariationModel(object):
|
||||
self.origLocations = locations
|
||||
self.axisOrder = axisOrder if axisOrder is not None else []
|
||||
self.extrapolate = extrapolate
|
||||
self.axisRanges = self.computeAxisRanges(locations) if extrapolate else None
|
||||
|
||||
locations = [{k: v for k, v in loc.items() if v != 0.0} for loc in locations]
|
||||
keyFunc = self.getMasterLocationsSortKeyFunc(
|
||||
@ -265,6 +275,17 @@ class VariationModel(object):
|
||||
self._subModels[key] = subModel
|
||||
return subModel, subList(key, items)
|
||||
|
||||
@staticmethod
|
||||
def computeAxisRanges(locations):
|
||||
axisRanges = {}
|
||||
allAxes = {axis for loc in locations for axis in loc.keys()}
|
||||
for loc in locations:
|
||||
for axis in allAxes:
|
||||
value = loc.get(axis, 0)
|
||||
axisMin, axisMax = axisRanges.get(axis, (value, value))
|
||||
axisRanges[axis] = min(value, axisMin), max(value, axisMax)
|
||||
return axisRanges
|
||||
|
||||
@staticmethod
|
||||
def getMasterLocationsSortKeyFunc(locations, axisOrder=[]):
|
||||
if {} not in locations:
|
||||
@ -439,8 +460,12 @@ class VariationModel(object):
|
||||
return model.getDeltas(items, round=round), model.supports
|
||||
|
||||
def getScalars(self, loc):
|
||||
return [supportScalar(loc, support, extrapolate=self.extrapolate)
|
||||
for support in self.supports]
|
||||
return [
|
||||
supportScalar(
|
||||
loc, support, extrapolate=self.extrapolate, axisRanges=self.axisRanges
|
||||
)
|
||||
for support in self.supports
|
||||
]
|
||||
|
||||
@staticmethod
|
||||
def interpolateFromDeltasAndScalars(deltas, scalars):
|
||||
|
||||
@ -36,10 +36,40 @@ def test_supportScalar():
|
||||
assert supportScalar({"wght": 0.2}, {}) == 1.0
|
||||
assert supportScalar({"wght": 0.2}, {"wght": (0, 2, 3)}) == 0.1
|
||||
assert supportScalar({"wght": 2.5}, {"wght": (0, 2, 4)}) == 0.75
|
||||
assert supportScalar({"wght": 4}, {"wght": (0, 2, 2)}) == 0.0
|
||||
assert supportScalar({"wght": 4}, {"wght": (0, 2, 2)}, extrapolate=True) == 2.0
|
||||
assert supportScalar({"wght": 4}, {"wght": (0, 2, 3)}, extrapolate=True) == 2.0
|
||||
assert supportScalar({"wght": 2}, {"wght": (0, 0.75, 1)}, extrapolate=True) == -4.0
|
||||
assert supportScalar({"wght": 3}, {"wght": (0, 2, 2)}) == 0.0
|
||||
assert supportScalar({"wght": 3}, {"wght": (0, 2, 2)}, extrapolate=True, axisRanges={"wght": (0, 2)}) == 1.5
|
||||
assert supportScalar({"wght": -1}, {"wght": (0, 2, 2)}, extrapolate=True, axisRanges={"wght": (0, 2)}) == -0.5
|
||||
assert supportScalar({"wght": 3}, {"wght": (0, 1, 2)}, extrapolate=True, axisRanges={"wght": (0, 2)}) == -1.0
|
||||
assert supportScalar({"wght": -1}, {"wght": (0, 1, 2)}, extrapolate=True, axisRanges={"wght": (0, 2)}) == -1.0
|
||||
assert supportScalar({"wght": 2}, {"wght": (0, 0.75, 1)}, extrapolate=True, axisRanges={"wght": (0, 1)}) == -4.0
|
||||
with pytest.raises(TypeError):
|
||||
supportScalar({"wght": 2}, {"wght": (0, 0.75, 1)}, extrapolate=True, axisRanges=None)
|
||||
|
||||
|
||||
def test_model_extrapolate():
|
||||
locations = [{}, {"a": 1}, {"b": 1}, {"a": 1, "b": 1}]
|
||||
model = VariationModel(locations, extrapolate=True)
|
||||
masterValues = [100, 200, 300, 400]
|
||||
testLocsAndValues = [
|
||||
({"a": -1, "b": -1}, -200),
|
||||
({"a": -1, "b": 0}, 0),
|
||||
({"a": -1, "b": 1}, 200),
|
||||
({"a": -1, "b": 2}, 400),
|
||||
({"a": 0, "b": -1}, -100),
|
||||
({"a": 0, "b": 0}, 100),
|
||||
({"a": 0, "b": 1}, 300),
|
||||
({"a": 0, "b": 2}, 500),
|
||||
({"a": 1, "b": -1}, 0),
|
||||
({"a": 1, "b": 0}, 200),
|
||||
({"a": 1, "b": 1}, 400),
|
||||
({"a": 1, "b": 2}, 600),
|
||||
({"a": 2, "b": -1}, 100),
|
||||
({"a": 2, "b": 0}, 300),
|
||||
({"a": 2, "b": 1}, 500),
|
||||
({"a": 2, "b": 2}, 700),
|
||||
]
|
||||
for loc, expectedValue in testLocsAndValues:
|
||||
assert expectedValue == model.interpolateFromMasters(loc, masterValues)
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user