457 lines
13 KiB
Python
Raw Normal View History

"""Variation fonts interpolation models."""
from __future__ import print_function, division, absolute_import
from fontTools.misc.py23 import *
__all__ = ['nonNone', 'allNone', 'allEqual', 'allEqualTo', 'subList',
'normalizeValue', 'normalizeLocation',
'supportScalar',
'VariationModel']
def nonNone(lst):
return [l for l in lst if l is not None]
2018-11-09 09:44:34 -05:00
def allNone(lst):
return all(l is None for l in lst)
def allEqualTo(ref, lst, mapper=None):
if mapper is None:
return all(ref == item for item in lst)
else:
mapped = mapper(ref)
return all(mapped == mapper(item) for item in lst)
2018-11-08 23:41:09 -05:00
def allEqual(lst, mapper=None):
if not lst:
return True
it = iter(lst)
first = next(it)
return allEqualTo(first, it, mapper=mapper)
def subList(truth, lst):
assert len(truth) == len(lst)
return [l for l,t in zip(lst,truth) if t]
def normalizeValue(v, triple):
"""Normalizes value based on a min/default/max triple.
>>> normalizeValue(400, (100, 400, 900))
0.0
>>> normalizeValue(100, (100, 400, 900))
-1.0
>>> normalizeValue(650, (100, 400, 900))
0.5
"""
lower, default, upper = triple
assert lower <= default <= upper, "invalid axis values: %3.3f, %3.3f %3.3f"%(lower, default, upper)
v = max(min(v, upper), lower)
if v == default:
v = 0.
elif v < default:
v = (v - default) / (default - lower)
else:
v = (v - default) / (upper - default)
return v
def normalizeLocation(location, axes):
"""Normalizes location based on axis min/default/max values from axes.
>>> axes = {"wght": (100, 400, 900)}
>>> normalizeLocation({"wght": 400}, axes)
{'wght': 0.0}
>>> normalizeLocation({"wght": 100}, axes)
{'wght': -1.0}
>>> normalizeLocation({"wght": 900}, axes)
{'wght': 1.0}
>>> normalizeLocation({"wght": 650}, axes)
{'wght': 0.5}
>>> normalizeLocation({"wght": 1000}, axes)
{'wght': 1.0}
>>> normalizeLocation({"wght": 0}, axes)
{'wght': -1.0}
>>> axes = {"wght": (0, 0, 1000)}
>>> normalizeLocation({"wght": 0}, axes)
{'wght': 0.0}
>>> normalizeLocation({"wght": -1}, axes)
{'wght': 0.0}
>>> normalizeLocation({"wght": 1000}, axes)
{'wght': 1.0}
2016-11-29 14:42:20 +01:00
>>> normalizeLocation({"wght": 500}, axes)
{'wght': 0.5}
>>> normalizeLocation({"wght": 1001}, axes)
{'wght': 1.0}
>>> axes = {"wght": (0, 1000, 1000)}
>>> normalizeLocation({"wght": 0}, axes)
{'wght': -1.0}
>>> normalizeLocation({"wght": -1}, axes)
{'wght': -1.0}
2016-11-29 14:42:20 +01:00
>>> normalizeLocation({"wght": 500}, axes)
{'wght': -0.5}
>>> normalizeLocation({"wght": 1000}, axes)
{'wght': 0.0}
>>> normalizeLocation({"wght": 1001}, axes)
{'wght': 0.0}
"""
out = {}
for tag,triple in axes.items():
v = location.get(tag, triple[1])
out[tag] = normalizeValue(v, triple)
return out
def supportScalar(location, support, ot=True):
"""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.
>>> supportScalar({}, {})
1.0
>>> supportScalar({'wght':.2}, {})
1.0
>>> supportScalar({'wght':.2}, {'wght':(0,2,3)})
0.1
>>> supportScalar({'wght':2.5}, {'wght':(0,2,4)})
0.75
>>> supportScalar({'wght':2.5, 'wdth':0}, {'wght':(0,2,4), 'wdth':(-1,0,+1)})
0.75
>>> supportScalar({'wght':2.5, 'wdth':.5}, {'wght':(0,2,4), 'wdth':(-1,0,+1)}, ot=False)
0.375
>>> supportScalar({'wght':2.5, 'wdth':0}, {'wght':(0,2,4), 'wdth':(-1,0,+1)})
0.75
>>> supportScalar({'wght':2.5, 'wdth':.5}, {'wght':(0,2,4), 'wdth':(-1,0,+1)})
0.75
"""
scalar = 1.
for axis,(lower,peak,upper) in support.items():
if ot:
# OpenType-specific case handling
if peak == 0.:
continue
if lower > peak or peak > upper:
continue
if lower < 0. and upper > 0.:
continue
v = location.get(axis, 0.)
else:
assert axis in location
v = location[axis]
if v == peak:
continue
if v <= lower or upper <= v:
scalar = 0.
break;
if v < peak:
scalar *= (v - lower) / (peak - lower)
else: # v > peak
scalar *= (v - upper) / (peak - upper)
return scalar
class VariationModel(object):
"""
Locations must be in normalized space. Ie. base master
is at origin (0).
>>> from pprint import pprint
>>> locations = [ \
{'wght':100}, \
{'wght':-100}, \
{'wght':-180}, \
{'wdth':+.3}, \
{'wght':+120,'wdth':.3}, \
{'wght':+120,'wdth':.2}, \
{}, \
{'wght':+180,'wdth':.3}, \
{'wght':+180}, \
]
>>> model = VariationModel(locations, axisOrder=['wght'])
>>> pprint(model.locations)
[{},
{'wght': -100},
{'wght': -180},
{'wght': 100},
{'wght': 180},
{'wdth': 0.3},
{'wdth': 0.3, 'wght': 180},
{'wdth': 0.3, 'wght': 120},
{'wdth': 0.2, 'wght': 120}]
>>> pprint(model.deltaWeights)
[{},
{0: 1.0},
{0: 1.0},
{0: 1.0},
{0: 1.0},
{0: 1.0},
{0: 1.0, 4: 1.0, 5: 1.0},
[varLib] Tweak support-resolution algorithm Improve varLib model algorithm. This, basically means any varfont built that had an unusual master configuration will change when rebuilt. Here's a good test: a two-axis with 8 masters at unusual locations: 2-----------------5----------3 | | | 7 | | | | 6 | | | | | | | 0-----------4----------------1 Previously, the reach of master 3 (Black Extended) would have started from A=.4, ie, the A position of master 4. It now correctly starts from 0. Same thing with masters after it. Ie, master 5 gets a span on the A axis from 0 to 1, whereas before it was getting from .4 to 1. Previously, the on-axis masters always cut the space. They don't anymore. That's more consistent, and fixes the main issue Erik showed at TYPO Labs 2017. Same issue was also causing the reach of master 3 to be limited, which I think is the issue being discussed in the linked issue. Both should be fixed. It's hard to describe exactly what happened before / after. Best to read the actual support values: Before: Sorted locations: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0.0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 1.0, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.6, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.5, 1.0), 'B': (0.0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] After: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 Sorted locations: [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.6, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] TODO: We should add this as a test case. There's another improvement I want to make, but that's separate.
2018-03-26 20:31:06 -07:00
{0: 1.0, 3: 0.75, 4: 0.25, 5: 1.0, 6: 0.6666666666666666},
{0: 1.0,
3: 0.75,
4: 0.25,
5: 0.6666666666666667,
[varLib] Tweak support-resolution algorithm Improve varLib model algorithm. This, basically means any varfont built that had an unusual master configuration will change when rebuilt. Here's a good test: a two-axis with 8 masters at unusual locations: 2-----------------5----------3 | | | 7 | | | | 6 | | | | | | | 0-----------4----------------1 Previously, the reach of master 3 (Black Extended) would have started from A=.4, ie, the A position of master 4. It now correctly starts from 0. Same thing with masters after it. Ie, master 5 gets a span on the A axis from 0 to 1, whereas before it was getting from .4 to 1. Previously, the on-axis masters always cut the space. They don't anymore. That's more consistent, and fixes the main issue Erik showed at TYPO Labs 2017. Same issue was also causing the reach of master 3 to be limited, which I think is the issue being discussed in the linked issue. Both should be fixed. It's hard to describe exactly what happened before / after. Best to read the actual support values: Before: Sorted locations: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0.0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 1.0, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.6, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.5, 1.0), 'B': (0.0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] After: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 Sorted locations: [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.6, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] TODO: We should add this as a test case. There's another improvement I want to make, but that's separate.
2018-03-26 20:31:06 -07:00
6: 0.4444444444444445,
7: 0.6666666666666667}]
"""
def __init__(self, locations, axisOrder=None):
if len(set(tuple(sorted(l.items())) for l in locations)) != len(locations):
raise ValueError("locations must be unique")
self.origLocations = locations
self.axisOrder = axisOrder if axisOrder is not None else []
locations = [{k:v for k,v in loc.items() if v != 0.} for loc in locations]
keyFunc = self.getMasterLocationsSortKeyFunc(locations, axisOrder=self.axisOrder)
self.locations = sorted(locations, key=keyFunc)
# Mapping from user's master order to our master order
self.mapping = [self.locations.index(l) for l in locations]
self.reverseMapping = [locations.index(l) for l in self.locations]
self._computeMasterSupports(keyFunc.axisPoints)
self._subModels = {}
def getSubModel(self, items):
if None not in items:
return self, items
key = tuple(v is not None for v in items)
subModel = self._subModels.get(key)
if subModel is None:
subModel = VariationModel(subList(key, self.origLocations), self.axisOrder)
self._subModels[key] = subModel
return subModel, subList(key, items)
@staticmethod
def getMasterLocationsSortKeyFunc(locations, axisOrder=[]):
assert {} in locations, "Base master not found."
axisPoints = {}
for loc in locations:
if len(loc) != 1:
continue
axis = next(iter(loc))
value = loc[axis]
if axis not in axisPoints:
axisPoints[axis] = {0.}
assert value not in axisPoints[axis], (
'Value "%s" in axisPoints["%s"] --> %s' % (value, axis, axisPoints)
)
axisPoints[axis].add(value)
def getKey(axisPoints, axisOrder):
def sign(v):
return -1 if v < 0 else +1 if v > 0 else 0
def key(loc):
rank = len(loc)
onPointAxes = [axis for axis,value in loc.items() if value in axisPoints[axis]]
orderedAxes = [axis for axis in axisOrder if axis in loc]
orderedAxes.extend([axis for axis in sorted(loc.keys()) if axis not in axisOrder])
return (
rank, # First, order by increasing rank
-len(onPointAxes), # Next, by decreasing number of onPoint axes
tuple(axisOrder.index(axis) if axis in axisOrder else 0x10000 for axis in orderedAxes), # Next, by known axes
tuple(orderedAxes), # Next, by all axes
tuple(sign(loc[axis]) for axis in orderedAxes), # Next, by signs of axis values
tuple(abs(loc[axis]) for axis in orderedAxes), # Next, by absolute value of axis values
)
return key
ret = getKey(axisPoints, axisOrder)
ret.axisPoints = axisPoints
return ret
def reorderMasters(self, master_list, mapping):
# For changing the master data order without
# recomputing supports and deltaWeights.
new_list = [master_list[idx] for idx in mapping]
self.origLocations = [self.origLocations[idx] for idx in mapping]
2018-11-30 21:46:16 -05:00
locations = [{k:v for k,v in loc.items() if v != 0.}
for loc in self.origLocations]
self.mapping = [self.locations.index(l) for l in locations]
self.reverseMapping = [locations.index(l) for l in self.locations]
self._subModels = {}
return new_list
2018-11-30 21:46:16 -05:00
def _computeMasterSupports(self, axisPoints):
supports = []
deltaWeights = []
locations = self.locations
# Compute min/max across each axis, use it as total range.
# TODO Take this as input from outside?
minV = {}
maxV = {}
for l in locations:
for k,v in l.items():
minV[k] = min(v, minV.get(k, v))
maxV[k] = max(v, maxV.get(k, v))
for i,loc in enumerate(locations):
box = {}
for axis,locV in loc.items():
[varLib] Tweak support-resolution algorithm Improve varLib model algorithm. This, basically means any varfont built that had an unusual master configuration will change when rebuilt. Here's a good test: a two-axis with 8 masters at unusual locations: 2-----------------5----------3 | | | 7 | | | | 6 | | | | | | | 0-----------4----------------1 Previously, the reach of master 3 (Black Extended) would have started from A=.4, ie, the A position of master 4. It now correctly starts from 0. Same thing with masters after it. Ie, master 5 gets a span on the A axis from 0 to 1, whereas before it was getting from .4 to 1. Previously, the on-axis masters always cut the space. They don't anymore. That's more consistent, and fixes the main issue Erik showed at TYPO Labs 2017. Same issue was also causing the reach of master 3 to be limited, which I think is the issue being discussed in the linked issue. Both should be fixed. It's hard to describe exactly what happened before / after. Best to read the actual support values: Before: Sorted locations: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0.0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 1.0, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.6, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.5, 1.0), 'B': (0.0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] After: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 Sorted locations: [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.6, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] TODO: We should add this as a test case. There's another improvement I want to make, but that's separate.
2018-03-26 20:31:06 -07:00
if locV > 0:
box[axis] = (0, locV, maxV[axis])
[varLib] Tweak support-resolution algorithm Improve varLib model algorithm. This, basically means any varfont built that had an unusual master configuration will change when rebuilt. Here's a good test: a two-axis with 8 masters at unusual locations: 2-----------------5----------3 | | | 7 | | | | 6 | | | | | | | 0-----------4----------------1 Previously, the reach of master 3 (Black Extended) would have started from A=.4, ie, the A position of master 4. It now correctly starts from 0. Same thing with masters after it. Ie, master 5 gets a span on the A axis from 0 to 1, whereas before it was getting from .4 to 1. Previously, the on-axis masters always cut the space. They don't anymore. That's more consistent, and fixes the main issue Erik showed at TYPO Labs 2017. Same issue was also causing the reach of master 3 to be limited, which I think is the issue being discussed in the linked issue. Both should be fixed. It's hard to describe exactly what happened before / after. Best to read the actual support values: Before: Sorted locations: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0.0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 1.0, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.6, 1.0), 'B': (0.0, 1.0, 1.0)}, {'A': (0.4, 0.5, 1.0), 'B': (0.0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] After: $ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .4,0 .6,1 .5,.5 .7,.7 Sorted locations: [{}, {'A': 0.4}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.6, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.7, 'B': 0.7}] Supports: [{}, {'A': (0, 0.4, 1.0)}, {'A': (0.4, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.6, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0.5, 0.7, 1.0), 'B': (0.5, 0.7, 1.0)}] TODO: We should add this as a test case. There's another improvement I want to make, but that's separate.
2018-03-26 20:31:06 -07:00
else:
box[axis] = (minV[axis], locV, 0)
locAxes = set(loc.keys())
# Walk over previous masters now
for j,m in enumerate(locations[:i]):
# Master with extra axes do not participte
if not set(m.keys()).issubset(locAxes):
continue
# If it's NOT in the current box, it does not participate
relevant = True
for axis, (lower,peak,upper) in box.items():
if axis not in m or not (m[axis] == peak or lower < m[axis] < upper):
relevant = False
break
if not relevant:
continue
[varLib.models] Refine modeling one last time In a523a697164a4 I changed the model such that when computing master supports and we hit another master, cut the box in one dimension only, to avoid interferring with other master. The axis to cut against was chosen as axis that minimizes volume loss. That was fine in a sense, but missed a great amount of information resulting from relative positioning of the two masters. As such, it could not distinguish between the following two situations: behdad:fonttools 0 (master*)$ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .5,.5 .55,.75 Sorted locations: [{}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.55, 'B': 0.75}] Supports: [{}, {'A': (0, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0, 0.55, 1.0), 'B': (0.5, 0.75, 1.0)}] behdad:fonttools 0 (02616ab...)$ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .5,.5 .75,.55 Sorted locations: [{}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.75, 'B': 0.55}] Supports: [{}, {'A': (0, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0, 0.75, 1.0), 'B': (0.5, 0.55, 1.0)}] Check the last line. In both cases the box was cut against the second axis. After this change, we get: behdad:fonttools 0 (master)$ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .5,.5 .55,.75 Sorted locations: [{}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.55, 'B': 0.75}] Supports: [{}, {'A': (0, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0, 0.55, 1.0), 'B': (0.5, 0.75, 1.0)}] behdad:fonttools 0 (master)$ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .5,.5 .75,.55 Sorted locations: [{}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.75, 'B': 0.55}] Supports: [{}, {'A': (0, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0.5, 0.75, 1.0), 'B': (0, 0.55, 1.0)}] Just draw the boxes on a piece of napkin and convince yourself that this is a good change. If the ratios are equal, we cut against the first axis in the axisOrder: behdad:fonttools 0 (master)$ ./fonttools varLib.models 0,0 0,1 1,0 1,1 .5,.5 .75,.75 Sorted locations: [{}, {'A': 1.0}, {'B': 1.0}, {'A': 1.0, 'B': 1.0}, {'A': 0.5, 'B': 0.5}, {'A': 0.75, 'B': 0.75}] Supports: [{}, {'A': (0, 1.0, 1.0)}, {'B': (0, 1.0, 1.0)}, {'A': (0, 1.0, 1.0), 'B': (0, 1.0, 1.0)}, {'A': (0, 0.5, 1.0), 'B': (0, 0.5, 1.0)}, {'A': (0.5, 0.75, 1.0), 'B': (0, 0.75, 1.0)}]
2018-04-27 16:47:46 -07:00
# Split the box for new master; split in whatever direction
# that has largest range ratio.
#
# For symmetry, we actually cut across multiple axes
# if they have the largest, equal, ratio.
# https://github.com/fonttools/fonttools/commit/7ee81c8821671157968b097f3e55309a1faa511e#commitcomment-31054804
bestAxes = {}
bestRatio = -1
for axis in m.keys():
val = m[axis]
assert axis in box
lower,locV,upper = box[axis]
newLower, newUpper = lower, upper
if val < locV:
newLower = val
ratio = (val - locV) / (lower - locV)
elif locV < val:
newUpper = val
ratio = (val - locV) / (upper - locV)
else: # val == locV
# Can't split box in this direction.
continue
if ratio > bestRatio:
bestAxes = {}
bestRatio = ratio
if ratio == bestRatio:
bestAxes[axis] = (newLower, locV, newUpper)
for axis,triple in bestAxes.items ():
box[axis] = triple
supports.append(box)
deltaWeight = {}
# Walk over previous masters now, populate deltaWeight
for j,m in enumerate(locations[:i]):
scalar = supportScalar(loc, supports[j])
if scalar:
deltaWeight[j] = scalar
deltaWeights.append(deltaWeight)
self.supports = supports
self.deltaWeights = deltaWeights
def getDeltas(self, masterValues):
assert len(masterValues) == len(self.deltaWeights)
mapping = self.reverseMapping
out = []
for i,weights in enumerate(self.deltaWeights):
delta = masterValues[mapping[i]]
for j,weight in weights.items():
delta -= out[j] * weight
out.append(delta)
return out
def getDeltasAndSupports(self, items):
model, items = self.getSubModel(items)
return model.getDeltas(items), model.supports
def getScalars(self, loc):
return [supportScalar(loc, support) for support in self.supports]
@staticmethod
def interpolateFromDeltasAndScalars(deltas, scalars):
v = None
assert len(deltas) == len(scalars)
for i,(delta,scalar) in enumerate(zip(deltas, scalars)):
if not scalar: continue
contribution = delta * scalar
if v is None:
v = contribution
else:
v += contribution
return v
def interpolateFromDeltas(self, loc, deltas):
scalars = self.getScalars(loc)
return self.interpolateFromDeltasAndScalars(deltas, scalars)
def interpolateFromMasters(self, loc, masterValues):
deltas = self.getDeltas(masterValues)
return self.interpolateFromDeltas(loc, deltas)
def interpolateFromMastersAndScalars(self, masterValues, scalars):
deltas = self.getDeltas(masterValues)
return self.interpolateFromDeltasAndScalars(deltas, scalars)
def piecewiseLinearMap(v, mapping):
keys = mapping.keys()
if not keys:
return v
if v in keys:
return mapping[v]
k = min(keys)
if v < k:
return v + mapping[k] - k
k = max(keys)
if v > k:
return v + mapping[k] - k
# Interpolate
a = max(k for k in keys if k < v)
b = min(k for k in keys if k > v)
va = mapping[a]
vb = mapping[b]
return va + (vb - va) * (v - a) / (b - a)
def main(args):
from fontTools import configLogger
args = args[1:]
# TODO: allow user to configure logging via command-line options
configLogger(level="INFO")
if len(args) < 1:
print("usage: fonttools varLib.models source.designspace", file=sys.stderr)
print(" or")
print("usage: fonttools varLib.models location1 location2 ...", file=sys.stderr)
sys.exit(1)
from pprint import pprint
if len(args) == 1 and args[0].endswith('.designspace'):
from fontTools.designspaceLib import DesignSpaceDocument
doc = DesignSpaceDocument()
doc.read(args[0])
locs = [s.location for s in doc.sources]
print("Original locations:")
pprint(locs)
doc.normalize()
print("Normalized locations:")
locs = [s.location for s in doc.sources]
pprint(locs)
else:
axes = [chr(c) for c in range(ord('A'), ord('Z')+1)]
locs = [dict(zip(axes, (float(v) for v in s.split(',')))) for s in args]
model = VariationModel(locs)
print("Sorted locations:")
pprint(model.locations)
print("Supports:")
pprint(model.supports)
if __name__ == "__main__":
import doctest, sys
if len(sys.argv) > 1:
sys.exit(main(sys.argv))
sys.exit(doctest.testmod().failed)