subset.py revision b61d7ac5530bfe452dfdb834208990505a321689
1# Copyright 2013 Google, Inc. All Rights Reserved.
2#
3# Google Author(s): Behdad Esfahbod
4
5"""Python OpenType Layout Subsetter.
6
7Later grown into full OpenType subsetter, supporting all standard tables.
8"""
9
10import sys
11import struct
12import time
13import array
14
15import fontTools.ttLib
16import fontTools.ttLib.tables
17import fontTools.ttLib.tables.otTables
18import fontTools.cffLib
19import fontTools.misc.psCharStrings
20import fontTools.pens.basePen
21
22
23def _add_method(*clazzes):
24  """Returns a decorator function that adds a new method to one or
25  more classes."""
26  def wrapper(method):
27    for clazz in clazzes:
28      assert clazz.__name__ != 'DefaultTable', 'Oops, table class not found.'
29      assert not hasattr(clazz, method.func_name), \
30          "Oops, class '%s' has method '%s'." % (clazz.__name__,
31                                                 method.func_name)
32      setattr(clazz, method.func_name, method)
33    return None
34  return wrapper
35
36def _uniq_sort(l):
37  return sorted(set(l))
38
39def _set_update(s, *others):
40  # Jython's set.update only takes one other argument.
41  # Emulate real set.update...
42  for other in others:
43    s.update(other)
44
45
46@_add_method(fontTools.ttLib.tables.otTables.Coverage)
47def intersect(self, glyphs):
48  "Returns ascending list of matching coverage values."
49  return [i for i,g in enumerate(self.glyphs) if g in glyphs]
50
51@_add_method(fontTools.ttLib.tables.otTables.Coverage)
52def intersect_glyphs(self, glyphs):
53  "Returns set of intersecting glyphs."
54  return set(g for g in self.glyphs if g in glyphs)
55
56@_add_method(fontTools.ttLib.tables.otTables.Coverage)
57def subset(self, glyphs):
58  "Returns ascending list of remaining coverage values."
59  indices = self.intersect(glyphs)
60  self.glyphs = [g for g in self.glyphs if g in glyphs]
61  return indices
62
63@_add_method(fontTools.ttLib.tables.otTables.Coverage)
64def remap(self, coverage_map):
65  "Remaps coverage."
66  self.glyphs = [self.glyphs[i] for i in coverage_map]
67
68@_add_method(fontTools.ttLib.tables.otTables.ClassDef)
69def intersect(self, glyphs):
70  "Returns ascending list of matching class values."
71  return _uniq_sort(
72     ([0] if any(g not in self.classDefs for g in glyphs) else []) +
73      [v for g,v in self.classDefs.iteritems() if g in glyphs])
74
75@_add_method(fontTools.ttLib.tables.otTables.ClassDef)
76def intersect_class(self, glyphs, klass):
77  "Returns set of glyphs matching class."
78  if klass == 0:
79    return set(g for g in glyphs if g not in self.classDefs)
80  return set(g for g,v in self.classDefs.iteritems()
81              if v == klass and g in glyphs)
82
83@_add_method(fontTools.ttLib.tables.otTables.ClassDef)
84def subset(self, glyphs, remap=False):
85  "Returns ascending list of remaining classes."
86  self.classDefs = dict((g,v) for g,v in self.classDefs.iteritems() if g in glyphs)
87  # Note: while class 0 has the special meaning of "not matched",
88  # if no glyph will ever /not match/, we can optimize class 0 out too.
89  indices = _uniq_sort(
90     ([0] if any(g not in self.classDefs for g in glyphs) else []) +
91      self.classDefs.values())
92  if remap:
93    self.remap(indices)
94  return indices
95
96@_add_method(fontTools.ttLib.tables.otTables.ClassDef)
97def remap(self, class_map):
98  "Remaps classes."
99  self.classDefs = dict((g,class_map.index(v))
100                         for g,v in self.classDefs.iteritems())
101
102@_add_method(fontTools.ttLib.tables.otTables.SingleSubst)
103def closure_glyphs(self, s, cur_glyphs=None):
104  if cur_glyphs == None: cur_glyphs = s.glyphs
105  if self.Format in [1, 2]:
106    s.glyphs.update(v for g,v in self.mapping.iteritems() if g in cur_glyphs)
107  else:
108    assert 0, "unknown format: %s" % self.Format
109
110@_add_method(fontTools.ttLib.tables.otTables.SingleSubst)
111def subset_glyphs(self, s):
112  if self.Format in [1, 2]:
113    self.mapping = dict((g,v) for g,v in self.mapping.iteritems()
114                        if g in s.glyphs and v in s.glyphs)
115    return bool(self.mapping)
116  else:
117    assert 0, "unknown format: %s" % self.Format
118
119@_add_method(fontTools.ttLib.tables.otTables.MultipleSubst)
120def closure_glyphs(self, s, cur_glyphs=None):
121  if cur_glyphs == None: cur_glyphs = s.glyphs
122  if self.Format == 1:
123    indices = self.Coverage.intersect(cur_glyphs)
124    _set_update(s.glyphs, *(self.Sequence[i].Substitute for i in indices))
125  else:
126    assert 0, "unknown format: %s" % self.Format
127
128@_add_method(fontTools.ttLib.tables.otTables.MultipleSubst)
129def subset_glyphs(self, s):
130  if self.Format == 1:
131    indices = self.Coverage.subset(s.glyphs)
132    self.Sequence = [self.Sequence[i] for i in indices]
133    # Now drop rules generating glyphs we don't want
134    indices = [i for i,seq in enumerate(self.Sequence)
135         if all(sub in s.glyphs for sub in seq.Substitute)]
136    self.Sequence = [self.Sequence[i] for i in indices]
137    self.Coverage.remap(indices)
138    self.SequenceCount = len(self.Sequence)
139    return bool(self.SequenceCount)
140  else:
141    assert 0, "unknown format: %s" % self.Format
142
143@_add_method(fontTools.ttLib.tables.otTables.AlternateSubst)
144def closure_glyphs(self, s, cur_glyphs=None):
145  if cur_glyphs == None: cur_glyphs = s.glyphs
146  if self.Format == 1:
147    _set_update(s.glyphs, *(vlist for g,vlist in self.alternates.iteritems()
148                            if g in cur_glyphs))
149  else:
150    assert 0, "unknown format: %s" % self.Format
151
152@_add_method(fontTools.ttLib.tables.otTables.AlternateSubst)
153def subset_glyphs(self, s):
154  if self.Format == 1:
155    self.alternates = dict((g,vlist)
156                           for g,vlist in self.alternates.iteritems()
157                           if g in s.glyphs and
158                              all(v in s.glyphs for v in vlist))
159    return bool(self.alternates)
160  else:
161    assert 0, "unknown format: %s" % self.Format
162
163@_add_method(fontTools.ttLib.tables.otTables.LigatureSubst)
164def closure_glyphs(self, s, cur_glyphs=None):
165  if cur_glyphs == None: cur_glyphs = s.glyphs
166  if self.Format == 1:
167    _set_update(s.glyphs, *([seq.LigGlyph for seq in seqs
168                             if all(c in s.glyphs for c in seq.Component)]
169                            for g,seqs in self.ligatures.iteritems()
170                            if g in cur_glyphs))
171  else:
172    assert 0, "unknown format: %s" % self.Format
173
174@_add_method(fontTools.ttLib.tables.otTables.LigatureSubst)
175def subset_glyphs(self, s):
176  if self.Format == 1:
177    self.ligatures = dict((g,v) for g,v in self.ligatures.iteritems()
178                          if g in s.glyphs)
179    self.ligatures = dict((g,[seq for seq in seqs
180                              if seq.LigGlyph in s.glyphs and
181                                 all(c in s.glyphs for c in seq.Component)])
182                           for g,seqs in self.ligatures.iteritems())
183    self.ligatures = dict((g,v) for g,v in self.ligatures.iteritems() if v)
184    return bool(self.ligatures)
185  else:
186    assert 0, "unknown format: %s" % self.Format
187
188@_add_method(fontTools.ttLib.tables.otTables.ReverseChainSingleSubst)
189def closure_glyphs(self, s, cur_glyphs=None):
190  if cur_glyphs == None: cur_glyphs = s.glyphs
191  if self.Format == 1:
192    indices = self.Coverage.intersect(cur_glyphs)
193    if(not indices or
194        not all(c.intersect(s.glyphs)
195                 for c in self.LookAheadCoverage + self.BacktrackCoverage)):
196      return
197    s.glyphs.update(self.Substitute[i] for i in indices)
198  else:
199    assert 0, "unknown format: %s" % self.Format
200
201@_add_method(fontTools.ttLib.tables.otTables.ReverseChainSingleSubst)
202def subset_glyphs(self, s):
203  if self.Format == 1:
204    indices = self.Coverage.subset(s.glyphs)
205    self.Substitute = [self.Substitute[i] for i in indices]
206    # Now drop rules generating glyphs we don't want
207    indices = [i for i,sub in enumerate(self.Substitute)
208         if sub in s.glyphs]
209    self.Substitute = [self.Substitute[i] for i in indices]
210    self.Coverage.remap(indices)
211    self.GlyphCount = len(self.Substitute)
212    return bool(self.GlyphCount and
213                 all(c.subset(s.glyphs)
214                      for c in self.LookAheadCoverage+self.BacktrackCoverage))
215  else:
216    assert 0, "unknown format: %s" % self.Format
217
218@_add_method(fontTools.ttLib.tables.otTables.SinglePos)
219def subset_glyphs(self, s):
220  if self.Format == 1:
221    return len(self.Coverage.subset(s.glyphs))
222  elif self.Format == 2:
223    indices = self.Coverage.subset(s.glyphs)
224    self.Value = [self.Value[i] for i in indices]
225    self.ValueCount = len(self.Value)
226    return bool(self.ValueCount)
227  else:
228    assert 0, "unknown format: %s" % self.Format
229
230@_add_method(fontTools.ttLib.tables.otTables.SinglePos)
231def prune_post_subset(self, options):
232  if not options.hinting:
233    # Drop device tables
234    self.ValueFormat &= ~0x00F0
235  return True
236
237@_add_method(fontTools.ttLib.tables.otTables.PairPos)
238def subset_glyphs(self, s):
239  if self.Format == 1:
240    indices = self.Coverage.subset(s.glyphs)
241    self.PairSet = [self.PairSet[i] for i in indices]
242    for p in self.PairSet:
243      p.PairValueRecord = [r for r in p.PairValueRecord
244                           if r.SecondGlyph in s.glyphs]
245      p.PairValueCount = len(p.PairValueRecord)
246    self.PairSet = [p for p in self.PairSet if p.PairValueCount]
247    self.PairSetCount = len(self.PairSet)
248    return bool(self.PairSetCount)
249  elif self.Format == 2:
250    class1_map = self.ClassDef1.subset(s.glyphs, remap=True)
251    class2_map = self.ClassDef2.subset(s.glyphs, remap=True)
252    self.Class1Record = [self.Class1Record[i] for i in class1_map]
253    for c in self.Class1Record:
254      c.Class2Record = [c.Class2Record[i] for i in class2_map]
255    self.Class1Count = len(class1_map)
256    self.Class2Count = len(class2_map)
257    return bool(self.Class1Count and
258                 self.Class2Count and
259                 self.Coverage.subset(s.glyphs))
260  else:
261    assert 0, "unknown format: %s" % self.Format
262
263@_add_method(fontTools.ttLib.tables.otTables.PairPos)
264def prune_post_subset(self, options):
265  if not options.hinting:
266    # Drop device tables
267    self.ValueFormat1 &= ~0x00F0
268    self.ValueFormat2 &= ~0x00F0
269  return True
270
271@_add_method(fontTools.ttLib.tables.otTables.CursivePos)
272def subset_glyphs(self, s):
273  if self.Format == 1:
274    indices = self.Coverage.subset(s.glyphs)
275    self.EntryExitRecord = [self.EntryExitRecord[i] for i in indices]
276    self.EntryExitCount = len(self.EntryExitRecord)
277    return bool(self.EntryExitCount)
278  else:
279    assert 0, "unknown format: %s" % self.Format
280
281@_add_method(fontTools.ttLib.tables.otTables.Anchor)
282def prune_hints(self):
283  # Drop device tables / contour anchor point
284  self.Format = 1
285
286@_add_method(fontTools.ttLib.tables.otTables.CursivePos)
287def prune_post_subset(self, options):
288  if not options.hinting:
289    for rec in self.EntryExitRecord:
290      if rec.EntryAnchor: rec.EntryAnchor.prune_hints()
291      if rec.ExitAnchor: rec.ExitAnchor.prune_hints()
292  return True
293
294@_add_method(fontTools.ttLib.tables.otTables.MarkBasePos)
295def subset_glyphs(self, s):
296  if self.Format == 1:
297    mark_indices = self.MarkCoverage.subset(s.glyphs)
298    self.MarkArray.MarkRecord = [self.MarkArray.MarkRecord[i]
299                                 for i in mark_indices]
300    self.MarkArray.MarkCount = len(self.MarkArray.MarkRecord)
301    base_indices = self.BaseCoverage.subset(s.glyphs)
302    self.BaseArray.BaseRecord = [self.BaseArray.BaseRecord[i]
303                                 for i in base_indices]
304    self.BaseArray.BaseCount = len(self.BaseArray.BaseRecord)
305    # Prune empty classes
306    class_indices = _uniq_sort(v.Class for v in self.MarkArray.MarkRecord)
307    self.ClassCount = len(class_indices)
308    for m in self.MarkArray.MarkRecord:
309      m.Class = class_indices.index(m.Class)
310    for b in self.BaseArray.BaseRecord:
311      b.BaseAnchor = [b.BaseAnchor[i] for i in class_indices]
312    return bool(self.ClassCount and
313                 self.MarkArray.MarkCount and
314                 self.BaseArray.BaseCount)
315  else:
316    assert 0, "unknown format: %s" % self.Format
317
318@_add_method(fontTools.ttLib.tables.otTables.MarkBasePos)
319def prune_post_subset(self, options):
320    if not options.hinting:
321      for m in self.MarkArray.MarkRecord:
322        m.MarkAnchor.prune_hints()
323      for b in self.BaseArray.BaseRecord:
324        for a in b.BaseAnchor:
325          a.prune_hints()
326    return True
327
328@_add_method(fontTools.ttLib.tables.otTables.MarkLigPos)
329def subset_glyphs(self, s):
330  if self.Format == 1:
331    mark_indices = self.MarkCoverage.subset(s.glyphs)
332    self.MarkArray.MarkRecord = [self.MarkArray.MarkRecord[i]
333                                 for i in mark_indices]
334    self.MarkArray.MarkCount = len(self.MarkArray.MarkRecord)
335    ligature_indices = self.LigatureCoverage.subset(s.glyphs)
336    self.LigatureArray.LigatureAttach = [self.LigatureArray.LigatureAttach[i]
337                                         for i in ligature_indices]
338    self.LigatureArray.LigatureCount = len(self.LigatureArray.LigatureAttach)
339    # Prune empty classes
340    class_indices = _uniq_sort(v.Class for v in self.MarkArray.MarkRecord)
341    self.ClassCount = len(class_indices)
342    for m in self.MarkArray.MarkRecord:
343      m.Class = class_indices.index(m.Class)
344    for l in self.LigatureArray.LigatureAttach:
345      for c in l.ComponentRecord:
346        c.LigatureAnchor = [c.LigatureAnchor[i] for i in class_indices]
347    return bool(self.ClassCount and
348                 self.MarkArray.MarkCount and
349                 self.LigatureArray.LigatureCount)
350  else:
351    assert 0, "unknown format: %s" % self.Format
352
353@_add_method(fontTools.ttLib.tables.otTables.MarkLigPos)
354def prune_post_subset(self, options):
355    if not options.hinting:
356      for m in self.MarkArray.MarkRecord:
357        m.MarkAnchor.prune_hints()
358      for l in self.LigatureArray.LigatureAttach:
359        for c in l.ComponentRecord:
360          for a in c.LigatureAnchor:
361            a.prune_hints()
362    return True
363
364@_add_method(fontTools.ttLib.tables.otTables.MarkMarkPos)
365def subset_glyphs(self, s):
366  if self.Format == 1:
367    mark1_indices = self.Mark1Coverage.subset(s.glyphs)
368    self.Mark1Array.MarkRecord = [self.Mark1Array.MarkRecord[i]
369                                  for i in mark1_indices]
370    self.Mark1Array.MarkCount = len(self.Mark1Array.MarkRecord)
371    mark2_indices = self.Mark2Coverage.subset(s.glyphs)
372    self.Mark2Array.Mark2Record = [self.Mark2Array.Mark2Record[i]
373                                   for i in mark2_indices]
374    self.Mark2Array.MarkCount = len(self.Mark2Array.Mark2Record)
375    # Prune empty classes
376    class_indices = _uniq_sort(v.Class for v in self.Mark1Array.MarkRecord)
377    self.ClassCount = len(class_indices)
378    for m in self.Mark1Array.MarkRecord:
379      m.Class = class_indices.index(m.Class)
380    for b in self.Mark2Array.Mark2Record:
381      b.Mark2Anchor = [b.Mark2Anchor[i] for i in class_indices]
382    return bool(self.ClassCount and
383                 self.Mark1Array.MarkCount and
384                 self.Mark2Array.MarkCount)
385  else:
386    assert 0, "unknown format: %s" % self.Format
387
388@_add_method(fontTools.ttLib.tables.otTables.MarkMarkPos)
389def prune_post_subset(self, options):
390    if not options.hinting:
391      # Drop device tables or contour anchor point
392      for m in self.Mark1Array.MarkRecord:
393        m.MarkAnchor.prune_hints()
394      for b in self.Mark2Array.Mark2Record:
395        for m in b.Mark2Anchor:
396          m.prune_hints()
397    return True
398
399@_add_method(fontTools.ttLib.tables.otTables.SingleSubst,
400             fontTools.ttLib.tables.otTables.MultipleSubst,
401             fontTools.ttLib.tables.otTables.AlternateSubst,
402             fontTools.ttLib.tables.otTables.LigatureSubst,
403             fontTools.ttLib.tables.otTables.ReverseChainSingleSubst,
404             fontTools.ttLib.tables.otTables.SinglePos,
405             fontTools.ttLib.tables.otTables.PairPos,
406             fontTools.ttLib.tables.otTables.CursivePos,
407             fontTools.ttLib.tables.otTables.MarkBasePos,
408             fontTools.ttLib.tables.otTables.MarkLigPos,
409             fontTools.ttLib.tables.otTables.MarkMarkPos)
410def subset_lookups(self, lookup_indices):
411  pass
412
413@_add_method(fontTools.ttLib.tables.otTables.SingleSubst,
414             fontTools.ttLib.tables.otTables.MultipleSubst,
415             fontTools.ttLib.tables.otTables.AlternateSubst,
416             fontTools.ttLib.tables.otTables.LigatureSubst,
417             fontTools.ttLib.tables.otTables.ReverseChainSingleSubst,
418             fontTools.ttLib.tables.otTables.SinglePos,
419             fontTools.ttLib.tables.otTables.PairPos,
420             fontTools.ttLib.tables.otTables.CursivePos,
421             fontTools.ttLib.tables.otTables.MarkBasePos,
422             fontTools.ttLib.tables.otTables.MarkLigPos,
423             fontTools.ttLib.tables.otTables.MarkMarkPos)
424def collect_lookups(self):
425  return []
426
427@_add_method(fontTools.ttLib.tables.otTables.SingleSubst,
428             fontTools.ttLib.tables.otTables.MultipleSubst,
429             fontTools.ttLib.tables.otTables.AlternateSubst,
430             fontTools.ttLib.tables.otTables.LigatureSubst,
431             fontTools.ttLib.tables.otTables.ContextSubst,
432             fontTools.ttLib.tables.otTables.ChainContextSubst,
433             fontTools.ttLib.tables.otTables.ReverseChainSingleSubst,
434             fontTools.ttLib.tables.otTables.SinglePos,
435             fontTools.ttLib.tables.otTables.PairPos,
436             fontTools.ttLib.tables.otTables.CursivePos,
437             fontTools.ttLib.tables.otTables.MarkBasePos,
438             fontTools.ttLib.tables.otTables.MarkLigPos,
439             fontTools.ttLib.tables.otTables.MarkMarkPos,
440             fontTools.ttLib.tables.otTables.ContextPos,
441             fontTools.ttLib.tables.otTables.ChainContextPos)
442def prune_pre_subset(self, options):
443  return True
444
445@_add_method(fontTools.ttLib.tables.otTables.SingleSubst,
446             fontTools.ttLib.tables.otTables.MultipleSubst,
447             fontTools.ttLib.tables.otTables.AlternateSubst,
448             fontTools.ttLib.tables.otTables.LigatureSubst,
449             fontTools.ttLib.tables.otTables.ReverseChainSingleSubst,
450             fontTools.ttLib.tables.otTables.ContextSubst,
451             fontTools.ttLib.tables.otTables.ChainContextSubst,
452             fontTools.ttLib.tables.otTables.ContextPos,
453             fontTools.ttLib.tables.otTables.ChainContextPos)
454def prune_post_subset(self, options):
455  return True
456
457@_add_method(fontTools.ttLib.tables.otTables.SingleSubst,
458             fontTools.ttLib.tables.otTables.AlternateSubst,
459             fontTools.ttLib.tables.otTables.ReverseChainSingleSubst)
460def may_have_non_1to1(self):
461  return False
462
463@_add_method(fontTools.ttLib.tables.otTables.MultipleSubst,
464             fontTools.ttLib.tables.otTables.LigatureSubst,
465             fontTools.ttLib.tables.otTables.ContextSubst,
466             fontTools.ttLib.tables.otTables.ChainContextSubst)
467def may_have_non_1to1(self):
468  return True
469
470@_add_method(fontTools.ttLib.tables.otTables.ContextSubst,
471             fontTools.ttLib.tables.otTables.ChainContextSubst,
472             fontTools.ttLib.tables.otTables.ContextPos,
473             fontTools.ttLib.tables.otTables.ChainContextPos)
474def __classify_context(self):
475
476  class ContextHelper(object):
477    def __init__(self, klass, Format):
478      if klass.__name__.endswith('Subst'):
479        Typ = 'Sub'
480        Type = 'Subst'
481      else:
482        Typ = 'Pos'
483        Type = 'Pos'
484      if klass.__name__.startswith('Chain'):
485        Chain = 'Chain'
486      else:
487        Chain = ''
488      ChainTyp = Chain+Typ
489
490      self.Typ = Typ
491      self.Type = Type
492      self.Chain = Chain
493      self.ChainTyp = ChainTyp
494
495      self.LookupRecord = Type+'LookupRecord'
496
497      if Format == 1:
498        Coverage = lambda r: r.Coverage
499        ChainCoverage = lambda r: r.Coverage
500        ContextData = lambda r:(None,)
501        ChainContextData = lambda r:(None, None, None)
502        RuleData = lambda r:(r.Input,)
503        ChainRuleData = lambda r:(r.Backtrack, r.Input, r.LookAhead)
504        SetRuleData = None
505        ChainSetRuleData = None
506      elif Format == 2:
507        Coverage = lambda r: r.Coverage
508        ChainCoverage = lambda r: r.Coverage
509        ContextData = lambda r:(r.ClassDef,)
510        ChainContextData = lambda r:(r.LookAheadClassDef,
511                                      r.InputClassDef,
512                                      r.BacktrackClassDef)
513        RuleData = lambda r:(r.Class,)
514        ChainRuleData = lambda r:(r.LookAhead, r.Input, r.Backtrack)
515        def SetRuleData(r, d):(r.Class,) = d
516        def ChainSetRuleData(r, d):(r.LookAhead, r.Input, r.Backtrack) = d
517      elif Format == 3:
518        Coverage = lambda r: r.Coverage[0]
519        ChainCoverage = lambda r: r.InputCoverage[0]
520        ContextData = None
521        ChainContextData = None
522        RuleData = lambda r: r.Coverage
523        ChainRuleData = lambda r:(r.LookAheadCoverage +
524                                   r.InputCoverage +
525                                   r.BacktrackCoverage)
526        SetRuleData = None
527        ChainSetRuleData = None
528      else:
529        assert 0, "unknown format: %s" % Format
530
531      if Chain:
532        self.Coverage = ChainCoverage
533        self.ContextData = ChainContextData
534        self.RuleData = ChainRuleData
535        self.SetRuleData = ChainSetRuleData
536      else:
537        self.Coverage = Coverage
538        self.ContextData = ContextData
539        self.RuleData = RuleData
540        self.SetRuleData = SetRuleData
541
542      if Format == 1:
543        self.Rule = ChainTyp+'Rule'
544        self.RuleCount = ChainTyp+'RuleCount'
545        self.RuleSet = ChainTyp+'RuleSet'
546        self.RuleSetCount = ChainTyp+'RuleSetCount'
547        self.Intersect = lambda glyphs, c, r: [r] if r in glyphs else []
548      elif Format == 2:
549        self.Rule = ChainTyp+'ClassRule'
550        self.RuleCount = ChainTyp+'ClassRuleCount'
551        self.RuleSet = ChainTyp+'ClassSet'
552        self.RuleSetCount = ChainTyp+'ClassSetCount'
553        self.Intersect = lambda glyphs, c, r: c.intersect_class(glyphs, r)
554
555        self.ClassDef = 'InputClassDef' if Chain else 'ClassDef'
556        self.Input = 'Input' if Chain else 'Class'
557
558  if self.Format not in [1, 2, 3]:
559    return None  # Don't shoot the messenger; let it go
560  if not hasattr(self.__class__, "__ContextHelpers"):
561    self.__class__.__ContextHelpers = {}
562  if self.Format not in self.__class__.__ContextHelpers:
563    helper = ContextHelper(self.__class__, self.Format)
564    self.__class__.__ContextHelpers[self.Format] = helper
565  return self.__class__.__ContextHelpers[self.Format]
566
567@_add_method(fontTools.ttLib.tables.otTables.ContextSubst,
568             fontTools.ttLib.tables.otTables.ChainContextSubst)
569def closure_glyphs(self, s, cur_glyphs=None):
570  if cur_glyphs == None: cur_glyphs = s.glyphs
571  c = self.__classify_context()
572
573  indices = c.Coverage(self).intersect(s.glyphs)
574  if not indices:
575    return []
576  cur_glyphs = c.Coverage(self).intersect_glyphs(s.glyphs);
577
578  if self.Format == 1:
579    ContextData = c.ContextData(self)
580    rss = getattr(self, c.RuleSet)
581    for i in indices:
582      if not rss[i]: continue
583      for r in getattr(rss[i], c.Rule):
584        if not r: continue
585        if all(all(c.Intersect(s.glyphs, cd, k) for k in klist)
586          for cd,klist in zip(ContextData, c.RuleData(r))):
587          chaos = False
588          for ll in getattr(r, c.LookupRecord):
589            if not ll: continue
590            seqi = ll.SequenceIndex
591            if chaos:
592              pos_glyphs = s.glyphs
593            else:
594              if seqi == 0:
595                pos_glyphs = set([c.Coverage(self).glyphs[i]])
596              else:
597                pos_glyphs = set([r.Input[seqi - 1]])
598            lookup = s.table.LookupList.Lookup[ll.LookupListIndex]
599            chaos = chaos or lookup.may_have_non_1to1()
600            lookup.closure_glyphs(s, cur_glyphs=pos_glyphs)
601  elif self.Format == 2:
602    ClassDef = getattr(self, c.ClassDef)
603    indices = ClassDef.intersect(cur_glyphs)
604    ContextData = c.ContextData(self)
605    rss = getattr(self, c.RuleSet)
606    for i in indices:
607      if not rss[i]: continue
608      for r in getattr(rss[i], c.Rule):
609        if not r: continue
610        if all(all(c.Intersect(s.glyphs, cd, k) for k in klist)
611          for cd,klist in zip(ContextData, c.RuleData(r))):
612          chaos = False
613          for ll in getattr(r, c.LookupRecord):
614            if not ll: continue
615            seqi = ll.SequenceIndex
616            if chaos:
617              pos_glyphs = s.glyphs
618            else:
619              if seqi == 0:
620                pos_glyphs = ClassDef.intersect_class(cur_glyphs, i)
621              else:
622                pos_glyphs = ClassDef.intersect_class(s.glyphs,
623                                                      getattr(r, c.Input)[seqi - 1])
624            lookup = s.table.LookupList.Lookup[ll.LookupListIndex]
625            chaos = chaos or lookup.may_have_non_1to1()
626            lookup.closure_glyphs(s, cur_glyphs=pos_glyphs)
627  elif self.Format == 3:
628    if not all(x.intersect(s.glyphs) for x in c.RuleData(self)):
629      return []
630    r = self
631    chaos = False
632    for ll in getattr(r, c.LookupRecord):
633      if not ll: continue
634      seqi = ll.SequenceIndex
635      if chaos:
636        pos_glyphs = s.glyphs
637      else:
638        if seqi == 0:
639          pos_glyphs = cur_glyphs
640        else:
641          pos_glyphs = r.InputCoverage[seqi].intersect_glyphs(s.glyphs)
642      lookup = s.table.LookupList.Lookup[ll.LookupListIndex]
643      chaos = chaos or lookup.may_have_non_1to1()
644      lookup.closure_glyphs(s, cur_glyphs=pos_glyphs)
645  else:
646    assert 0, "unknown format: %s" % self.Format
647
648@_add_method(fontTools.ttLib.tables.otTables.ContextSubst,
649             fontTools.ttLib.tables.otTables.ContextPos,
650             fontTools.ttLib.tables.otTables.ChainContextSubst,
651             fontTools.ttLib.tables.otTables.ChainContextPos)
652def subset_glyphs(self, s):
653  c = self.__classify_context()
654
655  if self.Format == 1:
656    indices = self.Coverage.subset(s.glyphs)
657    rss = getattr(self, c.RuleSet)
658    rss = [rss[i] for i in indices]
659    for rs in rss:
660      if not rs: continue
661      ss = getattr(rs, c.Rule)
662      ss = [r for r in ss
663            if r and all(all(g in s.glyphs for g in glist)
664              for glist in c.RuleData(r))]
665      setattr(rs, c.Rule, ss)
666      setattr(rs, c.RuleCount, len(ss))
667    # Prune empty subrulesets
668    rss = [rs for rs in rss if rs and getattr(rs, c.Rule)]
669    setattr(self, c.RuleSet, rss)
670    setattr(self, c.RuleSetCount, len(rss))
671    return bool(rss)
672  elif self.Format == 2:
673    if not self.Coverage.subset(s.glyphs):
674      return False
675    indices = getattr(self, c.ClassDef).subset(self.Coverage.glyphs,
676                                                 remap=False)
677    rss = getattr(self, c.RuleSet)
678    rss = [rss[i] for i in indices]
679    ContextData = c.ContextData(self)
680    klass_maps = [x.subset(s.glyphs, remap=True) for x in ContextData]
681    for rs in rss:
682      if not rs: continue
683      ss = getattr(rs, c.Rule)
684      ss = [r for r in ss
685            if r and all(all(k in klass_map for k in klist)
686              for klass_map,klist in zip(klass_maps, c.RuleData(r)))]
687      setattr(rs, c.Rule, ss)
688      setattr(rs, c.RuleCount, len(ss))
689
690      # Remap rule classes
691      for r in ss:
692        c.SetRuleData(r, [[klass_map.index(k) for k in klist]
693               for klass_map,klist in zip(klass_maps, c.RuleData(r))])
694    # Prune empty subrulesets
695    rss = [rs for rs in rss if rs and getattr(rs, c.Rule)]
696    setattr(self, c.RuleSet, rss)
697    setattr(self, c.RuleSetCount, len(rss))
698    return bool(rss)
699  elif self.Format == 3:
700    return all(x.subset(s.glyphs) for x in c.RuleData(self))
701  else:
702    assert 0, "unknown format: %s" % self.Format
703
704@_add_method(fontTools.ttLib.tables.otTables.ContextSubst,
705             fontTools.ttLib.tables.otTables.ChainContextSubst,
706             fontTools.ttLib.tables.otTables.ContextPos,
707             fontTools.ttLib.tables.otTables.ChainContextPos)
708def subset_lookups(self, lookup_indices):
709  c = self.__classify_context()
710
711  if self.Format in [1, 2]:
712    for rs in getattr(self, c.RuleSet):
713      if not rs: continue
714      for r in getattr(rs, c.Rule):
715        if not r: continue
716        setattr(r, c.LookupRecord,
717                 [ll for ll in getattr(r, c.LookupRecord)
718                  if ll and ll.LookupListIndex in lookup_indices])
719        for ll in getattr(r, c.LookupRecord):
720          if not ll: continue
721          ll.LookupListIndex = lookup_indices.index(ll.LookupListIndex)
722  elif self.Format == 3:
723    setattr(self, c.LookupRecord,
724             [ll for ll in getattr(self, c.LookupRecord)
725              if ll and ll.LookupListIndex in lookup_indices])
726    for ll in getattr(self, c.LookupRecord):
727      if not ll: continue
728      ll.LookupListIndex = lookup_indices.index(ll.LookupListIndex)
729  else:
730    assert 0, "unknown format: %s" % self.Format
731
732@_add_method(fontTools.ttLib.tables.otTables.ContextSubst,
733             fontTools.ttLib.tables.otTables.ChainContextSubst,
734             fontTools.ttLib.tables.otTables.ContextPos,
735             fontTools.ttLib.tables.otTables.ChainContextPos)
736def collect_lookups(self):
737  c = self.__classify_context()
738
739  if self.Format in [1, 2]:
740    return [ll.LookupListIndex
741      for rs in getattr(self, c.RuleSet) if rs
742      for r in getattr(rs, c.Rule) if r
743      for ll in getattr(r, c.LookupRecord) if ll]
744  elif self.Format == 3:
745    return [ll.LookupListIndex
746      for ll in getattr(self, c.LookupRecord) if ll]
747  else:
748    assert 0, "unknown format: %s" % self.Format
749
750@_add_method(fontTools.ttLib.tables.otTables.ExtensionSubst)
751def closure_glyphs(self, s, cur_glyphs=None):
752  if self.Format == 1:
753    self.ExtSubTable.closure_glyphs(s, cur_glyphs)
754  else:
755    assert 0, "unknown format: %s" % self.Format
756
757@_add_method(fontTools.ttLib.tables.otTables.ExtensionSubst)
758def may_have_non_1to1(self):
759  if self.Format == 1:
760    return self.ExtSubTable.may_have_non_1to1()
761  else:
762    assert 0, "unknown format: %s" % self.Format
763
764@_add_method(fontTools.ttLib.tables.otTables.ExtensionSubst,
765             fontTools.ttLib.tables.otTables.ExtensionPos)
766def prune_pre_subset(self, options):
767  if self.Format == 1:
768    return self.ExtSubTable.prune_pre_subset(options)
769  else:
770    assert 0, "unknown format: %s" % self.Format
771
772@_add_method(fontTools.ttLib.tables.otTables.ExtensionSubst,
773             fontTools.ttLib.tables.otTables.ExtensionPos)
774def subset_glyphs(self, s):
775  if self.Format == 1:
776    return self.ExtSubTable.subset_glyphs(s)
777  else:
778    assert 0, "unknown format: %s" % self.Format
779
780@_add_method(fontTools.ttLib.tables.otTables.ExtensionSubst,
781             fontTools.ttLib.tables.otTables.ExtensionPos)
782def prune_post_subset(self, options):
783  if self.Format == 1:
784    return self.ExtSubTable.prune_post_subset(options)
785  else:
786    assert 0, "unknown format: %s" % self.Format
787
788@_add_method(fontTools.ttLib.tables.otTables.ExtensionSubst,
789             fontTools.ttLib.tables.otTables.ExtensionPos)
790def subset_lookups(self, lookup_indices):
791  if self.Format == 1:
792    return self.ExtSubTable.subset_lookups(lookup_indices)
793  else:
794    assert 0, "unknown format: %s" % self.Format
795
796@_add_method(fontTools.ttLib.tables.otTables.ExtensionSubst,
797             fontTools.ttLib.tables.otTables.ExtensionPos)
798def collect_lookups(self):
799  if self.Format == 1:
800    return self.ExtSubTable.collect_lookups()
801  else:
802    assert 0, "unknown format: %s" % self.Format
803
804@_add_method(fontTools.ttLib.tables.otTables.Lookup)
805def closure_glyphs(self, s, cur_glyphs=None):
806  for st in self.SubTable:
807    if not st: continue
808    st.closure_glyphs(s, cur_glyphs)
809
810@_add_method(fontTools.ttLib.tables.otTables.Lookup)
811def prune_pre_subset(self, options):
812  ret = False
813  for st in self.SubTable:
814    if not st: continue
815    if st.prune_pre_subset(options): ret = True
816  return ret
817
818@_add_method(fontTools.ttLib.tables.otTables.Lookup)
819def subset_glyphs(self, s):
820  self.SubTable = [st for st in self.SubTable if st and st.subset_glyphs(s)]
821  self.SubTableCount = len(self.SubTable)
822  return bool(self.SubTableCount)
823
824@_add_method(fontTools.ttLib.tables.otTables.Lookup)
825def prune_post_subset(self, options):
826  ret = False
827  for st in self.SubTable:
828    if not st: continue
829    if st.prune_post_subset(options): ret = True
830  return ret
831
832@_add_method(fontTools.ttLib.tables.otTables.Lookup)
833def subset_lookups(self, lookup_indices):
834  for s in self.SubTable:
835    s.subset_lookups(lookup_indices)
836
837@_add_method(fontTools.ttLib.tables.otTables.Lookup)
838def collect_lookups(self):
839  return _uniq_sort(sum((st.collect_lookups() for st in self.SubTable
840                         if st), []))
841
842@_add_method(fontTools.ttLib.tables.otTables.Lookup)
843def may_have_non_1to1(self):
844  return any(st.may_have_non_1to1() for st in self.SubTable if st)
845
846@_add_method(fontTools.ttLib.tables.otTables.LookupList)
847def prune_pre_subset(self, options):
848  ret = False
849  for l in self.Lookup:
850    if not l: continue
851    if l.prune_pre_subset(options): ret = True
852  return ret
853
854@_add_method(fontTools.ttLib.tables.otTables.LookupList)
855def subset_glyphs(self, s):
856  "Returns the indices of nonempty lookups."
857  return [i for i,l in enumerate(self.Lookup) if l and l.subset_glyphs(s)]
858
859@_add_method(fontTools.ttLib.tables.otTables.LookupList)
860def prune_post_subset(self, options):
861  ret = False
862  for l in self.Lookup:
863    if not l: continue
864    if l.prune_post_subset(options): ret = True
865  return ret
866
867@_add_method(fontTools.ttLib.tables.otTables.LookupList)
868def subset_lookups(self, lookup_indices):
869  self.Lookup = [self.Lookup[i] for i in lookup_indices
870                 if i < self.LookupCount]
871  self.LookupCount = len(self.Lookup)
872  for l in self.Lookup:
873    l.subset_lookups(lookup_indices)
874
875@_add_method(fontTools.ttLib.tables.otTables.LookupList)
876def closure_lookups(self, lookup_indices):
877  lookup_indices = _uniq_sort(lookup_indices)
878  recurse = lookup_indices
879  while True:
880    recurse_lookups = sum((self.Lookup[i].collect_lookups()
881                            for i in recurse if i < self.LookupCount), [])
882    recurse_lookups = [l for l in recurse_lookups
883                       if l not in lookup_indices and l < self.LookupCount]
884    if not recurse_lookups:
885      return _uniq_sort(lookup_indices)
886    recurse_lookups = _uniq_sort(recurse_lookups)
887    lookup_indices.extend(recurse_lookups)
888    recurse = recurse_lookups
889
890@_add_method(fontTools.ttLib.tables.otTables.Feature)
891def subset_lookups(self, lookup_indices):
892  self.LookupListIndex = [l for l in self.LookupListIndex
893                          if l in lookup_indices]
894  # Now map them.
895  self.LookupListIndex = [lookup_indices.index(l)
896                          for l in self.LookupListIndex]
897  self.LookupCount = len(self.LookupListIndex)
898  return self.LookupCount
899
900@_add_method(fontTools.ttLib.tables.otTables.Feature)
901def collect_lookups(self):
902  return self.LookupListIndex[:]
903
904@_add_method(fontTools.ttLib.tables.otTables.FeatureList)
905def subset_lookups(self, lookup_indices):
906  "Returns the indices of nonempty features."
907  feature_indices = [i for i,f in enumerate(self.FeatureRecord)
908                     if f.Feature.subset_lookups(lookup_indices)]
909  self.subset_features(feature_indices)
910  return feature_indices
911
912@_add_method(fontTools.ttLib.tables.otTables.FeatureList)
913def collect_lookups(self, feature_indices):
914  return _uniq_sort(sum((self.FeatureRecord[i].Feature.collect_lookups()
915                         for i in feature_indices
916                          if i < self.FeatureCount), []))
917
918@_add_method(fontTools.ttLib.tables.otTables.FeatureList)
919def subset_features(self, feature_indices):
920  self.FeatureRecord = [self.FeatureRecord[i] for i in feature_indices]
921  self.FeatureCount = len(self.FeatureRecord)
922  return bool(self.FeatureCount)
923
924@_add_method(fontTools.ttLib.tables.otTables.DefaultLangSys,
925             fontTools.ttLib.tables.otTables.LangSys)
926def subset_features(self, feature_indices):
927  if self.ReqFeatureIndex in feature_indices:
928    self.ReqFeatureIndex = feature_indices.index(self.ReqFeatureIndex)
929  else:
930    self.ReqFeatureIndex = 65535
931  self.FeatureIndex = [f for f in self.FeatureIndex if f in feature_indices]
932  # Now map them.
933  self.FeatureIndex = [feature_indices.index(f) for f in self.FeatureIndex
934                       if f in feature_indices]
935  self.FeatureCount = len(self.FeatureIndex)
936  return bool(self.FeatureCount or self.ReqFeatureIndex != 65535)
937
938@_add_method(fontTools.ttLib.tables.otTables.DefaultLangSys,
939             fontTools.ttLib.tables.otTables.LangSys)
940def collect_features(self):
941  feature_indices = self.FeatureIndex[:]
942  if self.ReqFeatureIndex != 65535:
943    feature_indices.append(self.ReqFeatureIndex)
944  return _uniq_sort(feature_indices)
945
946@_add_method(fontTools.ttLib.tables.otTables.Script)
947def subset_features(self, feature_indices):
948  if(self.DefaultLangSys and
949      not self.DefaultLangSys.subset_features(feature_indices)):
950    self.DefaultLangSys = None
951  self.LangSysRecord = [l for l in self.LangSysRecord
952                        if l.LangSys.subset_features(feature_indices)]
953  self.LangSysCount = len(self.LangSysRecord)
954  return bool(self.LangSysCount or self.DefaultLangSys)
955
956@_add_method(fontTools.ttLib.tables.otTables.Script)
957def collect_features(self):
958  feature_indices = [l.LangSys.collect_features() for l in self.LangSysRecord]
959  if self.DefaultLangSys:
960    feature_indices.append(self.DefaultLangSys.collect_features())
961  return _uniq_sort(sum(feature_indices, []))
962
963@_add_method(fontTools.ttLib.tables.otTables.ScriptList)
964def subset_features(self, feature_indices):
965  self.ScriptRecord = [s for s in self.ScriptRecord
966                       if s.Script.subset_features(feature_indices)]
967  self.ScriptCount = len(self.ScriptRecord)
968  return bool(self.ScriptCount)
969
970@_add_method(fontTools.ttLib.tables.otTables.ScriptList)
971def collect_features(self):
972  return _uniq_sort(sum((s.Script.collect_features()
973                         for s in self.ScriptRecord), []))
974
975@_add_method(fontTools.ttLib.getTableClass('GSUB'))
976def closure_glyphs(self, s):
977  s.table = self.table
978  feature_indices = self.table.ScriptList.collect_features()
979  lookup_indices = self.table.FeatureList.collect_lookups(feature_indices)
980  while True:
981    orig_glyphs = s.glyphs.copy()
982    for i in lookup_indices:
983      if i >= self.table.LookupList.LookupCount: continue
984      if not self.table.LookupList.Lookup[i]: continue
985      self.table.LookupList.Lookup[i].closure_glyphs(s)
986    if orig_glyphs == s.glyphs:
987      break
988  del s.table
989
990@_add_method(fontTools.ttLib.getTableClass('GSUB'),
991             fontTools.ttLib.getTableClass('GPOS'))
992def subset_glyphs(self, s):
993  s.glyphs = s.glyphs_gsubed
994  lookup_indices = self.table.LookupList.subset_glyphs(s)
995  self.subset_lookups(lookup_indices)
996  self.prune_lookups()
997  return True
998
999@_add_method(fontTools.ttLib.getTableClass('GSUB'),
1000             fontTools.ttLib.getTableClass('GPOS'))
1001def subset_lookups(self, lookup_indices):
1002  """Retrains specified lookups, then removes empty features, language
1003     systems, and scripts."""
1004  self.table.LookupList.subset_lookups(lookup_indices)
1005  feature_indices = self.table.FeatureList.subset_lookups(lookup_indices)
1006  self.table.ScriptList.subset_features(feature_indices)
1007
1008@_add_method(fontTools.ttLib.getTableClass('GSUB'),
1009             fontTools.ttLib.getTableClass('GPOS'))
1010def prune_lookups(self):
1011  "Remove unreferenced lookups"
1012  feature_indices = self.table.ScriptList.collect_features()
1013  lookup_indices = self.table.FeatureList.collect_lookups(feature_indices)
1014  lookup_indices = self.table.LookupList.closure_lookups(lookup_indices)
1015  self.subset_lookups(lookup_indices)
1016
1017@_add_method(fontTools.ttLib.getTableClass('GSUB'),
1018             fontTools.ttLib.getTableClass('GPOS'))
1019def subset_feature_tags(self, feature_tags):
1020  feature_indices = [i for i,f in
1021                     enumerate(self.table.FeatureList.FeatureRecord)
1022                     if f.FeatureTag in feature_tags]
1023  self.table.FeatureList.subset_features(feature_indices)
1024  self.table.ScriptList.subset_features(feature_indices)
1025
1026@_add_method(fontTools.ttLib.getTableClass('GSUB'),
1027             fontTools.ttLib.getTableClass('GPOS'))
1028def prune_pre_subset(self, options):
1029  if '*' not in options.layout_features:
1030    self.subset_feature_tags(options.layout_features)
1031  self.prune_lookups()
1032  self.table.LookupList.prune_pre_subset(options);
1033  return True
1034
1035@_add_method(fontTools.ttLib.getTableClass('GSUB'),
1036             fontTools.ttLib.getTableClass('GPOS'))
1037def prune_post_subset(self, options):
1038  self.table.LookupList.prune_post_subset(options);
1039  return True
1040
1041@_add_method(fontTools.ttLib.getTableClass('GDEF'))
1042def subset_glyphs(self, s):
1043  glyphs = s.glyphs_gsubed
1044  table = self.table
1045  if table.LigCaretList:
1046    indices = table.LigCaretList.Coverage.subset(glyphs)
1047    table.LigCaretList.LigGlyph = [table.LigCaretList.LigGlyph[i]
1048                                   for i in indices]
1049    table.LigCaretList.LigGlyphCount = len(table.LigCaretList.LigGlyph)
1050    if not table.LigCaretList.LigGlyphCount:
1051      table.LigCaretList = None
1052  if table.MarkAttachClassDef:
1053    table.MarkAttachClassDef.classDefs = dict((g,v) for g,v in
1054                                              table.MarkAttachClassDef.
1055                                                classDefs.iteritems()
1056                                              if g in glyphs)
1057    if not table.MarkAttachClassDef.classDefs:
1058      table.MarkAttachClassDef = None
1059  if table.GlyphClassDef:
1060    table.GlyphClassDef.classDefs = dict((g,v) for g,v in
1061                                         table.GlyphClassDef.
1062                                           classDefs.iteritems()
1063                                         if g in glyphs)
1064    if not table.GlyphClassDef.classDefs:
1065      table.GlyphClassDef = None
1066  if table.AttachList:
1067    indices = table.AttachList.Coverage.subset(glyphs)
1068    table.AttachList.AttachPoint = [table.AttachList.AttachPoint[i]
1069                                    for i in indices]
1070    table.AttachList.GlyphCount = len(table.AttachList.AttachPoint)
1071    if not table.AttachList.GlyphCount:
1072      table.AttachList = None
1073  return bool(table.LigCaretList or
1074               table.MarkAttachClassDef or
1075               table.GlyphClassDef or
1076               table.AttachList)
1077
1078@_add_method(fontTools.ttLib.getTableClass('kern'))
1079def prune_pre_subset(self, options):
1080  # Prune unknown kern table types
1081  self.kernTables = [t for t in self.kernTables if hasattr(t, 'kernTable')]
1082  return bool(self.kernTables)
1083
1084@_add_method(fontTools.ttLib.getTableClass('kern'))
1085def subset_glyphs(self, s):
1086  glyphs = s.glyphs_gsubed
1087  for t in self.kernTables:
1088    t.kernTable = dict(((a,b),v) for (a,b),v in t.kernTable.iteritems()
1089                       if a in glyphs and b in glyphs)
1090  self.kernTables = [t for t in self.kernTables if t.kernTable]
1091  return bool(self.kernTables)
1092
1093@_add_method(fontTools.ttLib.getTableClass('vmtx'),
1094             fontTools.ttLib.getTableClass('hmtx'))
1095def subset_glyphs(self, s):
1096  self.metrics = dict((g,v) for g,v in self.metrics.iteritems() if g in s.glyphs)
1097  return bool(self.metrics)
1098
1099@_add_method(fontTools.ttLib.getTableClass('hdmx'))
1100def subset_glyphs(self, s):
1101  self.hdmx = dict((sz,_dict((g,v) for g,v in l.iteritems() if g in s.glyphs))
1102                   for sz,l in self.hdmx.iteritems())
1103  return bool(self.hdmx)
1104
1105@_add_method(fontTools.ttLib.getTableClass('VORG'))
1106def subset_glyphs(self, s):
1107  self.VOriginRecords = dict((g,v) for g,v in self.VOriginRecords.iteritems()
1108                             if g in s.glyphs)
1109  self.numVertOriginYMetrics = len(self.VOriginRecords)
1110  return True  # Never drop; has default metrics
1111
1112@_add_method(fontTools.ttLib.getTableClass('post'))
1113def prune_pre_subset(self, options):
1114  if not options.glyph_names:
1115    self.formatType = 3.0
1116  return True
1117
1118@_add_method(fontTools.ttLib.getTableClass('post'))
1119def subset_glyphs(self, s):
1120  self.extraNames = []  # This seems to do it
1121  return True
1122
1123@_add_method(fontTools.ttLib.getTableModule('glyf').Glyph)
1124def getComponentNamesFast(self, glyfTable):
1125  if not self.data or struct.unpack(">h", self.data[:2])[0] >= 0:
1126    return []  # Not composite
1127  data = self.data
1128  i = 10
1129  components = []
1130  more = 1
1131  while more:
1132    flags, glyphID = struct.unpack(">HH", data[i:i+4])
1133    i += 4
1134    flags = int(flags)
1135    components.append(glyfTable.getGlyphName(int(glyphID)))
1136
1137    if flags & 0x0001: i += 4  # ARG_1_AND_2_ARE_WORDS
1138    else: i += 2
1139    if flags & 0x0008: i += 2  # WE_HAVE_A_SCALE
1140    elif flags & 0x0040: i += 4  # WE_HAVE_AN_X_AND_Y_SCALE
1141    elif flags & 0x0080: i += 8  # WE_HAVE_A_TWO_BY_TWO
1142    more = flags & 0x0020  # MORE_COMPONENTS
1143
1144  return components
1145
1146@_add_method(fontTools.ttLib.getTableModule('glyf').Glyph)
1147def remapComponentsFast(self, indices):
1148  if not self.data or struct.unpack(">h", self.data[:2])[0] >= 0:
1149    return  # Not composite
1150  data = array.array("B", self.data)
1151  i = 10
1152  more = 1
1153  while more:
1154    flags =(data[i] << 8) | data[i+1]
1155    glyphID =(data[i+2] << 8) | data[i+3]
1156    # Remap
1157    glyphID = indices.index(glyphID)
1158    data[i+2] = glyphID >> 8
1159    data[i+3] = glyphID & 0xFF
1160    i += 4
1161    flags = int(flags)
1162
1163    if flags & 0x0001: i += 4  # ARG_1_AND_2_ARE_WORDS
1164    else: i += 2
1165    if flags & 0x0008: i += 2  # WE_HAVE_A_SCALE
1166    elif flags & 0x0040: i += 4  # WE_HAVE_AN_X_AND_Y_SCALE
1167    elif flags & 0x0080: i += 8  # WE_HAVE_A_TWO_BY_TWO
1168    more = flags & 0x0020  # MORE_COMPONENTS
1169
1170  self.data = data.tostring()
1171
1172@_add_method(fontTools.ttLib.getTableModule('glyf').Glyph)
1173def dropInstructionsFast(self):
1174  if not self.data:
1175    return
1176  numContours = struct.unpack(">h", self.data[:2])[0]
1177  data = array.array("B", self.data)
1178  i = 10
1179  if numContours >= 0:
1180    i += 2 * numContours  # endPtsOfContours
1181    instructionLen =(data[i] << 8) | data[i+1]
1182    # Zero it
1183    data[i] = data [i+1] = 0
1184    i += 2
1185    if instructionLen:
1186      # Splice it out
1187      data = data[:i] + data[i+instructionLen:]
1188  else:
1189    more = 1
1190    while more:
1191      flags =(data[i] << 8) | data[i+1]
1192      # Turn instruction flag off
1193      flags &= ~0x0100  # WE_HAVE_INSTRUCTIONS
1194      data[i+0] = flags >> 8
1195      data[i+1] = flags & 0xFF
1196      i += 4
1197      flags = int(flags)
1198
1199      if flags & 0x0001: i += 4  # ARG_1_AND_2_ARE_WORDS
1200      else: i += 2
1201      if flags & 0x0008: i += 2  # WE_HAVE_A_SCALE
1202      elif flags & 0x0040: i += 4  # WE_HAVE_AN_X_AND_Y_SCALE
1203      elif flags & 0x0080: i += 8  # WE_HAVE_A_TWO_BY_TWO
1204      more = flags & 0x0020  # MORE_COMPONENTS
1205
1206    # Cut off
1207    data = data[:i]
1208  if len(data) % 4:
1209    # add pad bytes
1210    nPadBytes = 4 -(len(data) % 4)
1211    for i in range(nPadBytes):
1212      data.append(0)
1213  self.data = data.tostring()
1214
1215@_add_method(fontTools.ttLib.getTableClass('glyf'))
1216def closure_glyphs(self, s):
1217  decompose = s.glyphs
1218  # I don't know if component glyphs can be composite themselves.
1219  # We handle them anyway.
1220  while True:
1221    components = set()
1222    for g in decompose:
1223      if g not in self.glyphs:
1224        continue
1225      gl = self.glyphs[g]
1226      if hasattr(gl, "data"):
1227        for c in gl.getComponentNamesFast(self):
1228          if c not in s.glyphs:
1229            components.add(c)
1230      else:
1231        # TTX seems to expand gid0..3 always
1232        if gl.isComposite():
1233          for c in gl.components:
1234            if c.glyphName not in s.glyphs:
1235              components.add(c.glyphName)
1236    components = set(c for c in components if c not in s.glyphs)
1237    if not components:
1238      break
1239    decompose = components
1240    s.glyphs.update(components)
1241
1242@_add_method(fontTools.ttLib.getTableClass('glyf'))
1243def prune_pre_subset(self, options):
1244  if options.notdef_glyph and not options.notdef_outline:
1245    g = self[self.glyphOrder[0]]
1246    # Yay, easy!
1247    g.__dict__.clear()
1248    g.data = ""
1249  return True
1250
1251@_add_method(fontTools.ttLib.getTableClass('glyf'))
1252def subset_glyphs(self, s):
1253  self.glyphs = dict((g,v) for g,v in self.glyphs.iteritems() if g in s.glyphs)
1254  indices = [i for i,g in enumerate(self.glyphOrder) if g in s.glyphs]
1255  for v in self.glyphs.itervalues():
1256    if hasattr(v, "data"):
1257      v.remapComponentsFast(indices)
1258    else:
1259      pass  # No need
1260  self.glyphOrder = [g for g in self.glyphOrder if g in s.glyphs]
1261  # Don't drop empty 'glyf' tables, otherwise 'loca' doesn't get subset.
1262  return True
1263
1264@_add_method(fontTools.ttLib.getTableClass('glyf'))
1265def prune_post_subset(self, options):
1266  if not options.hinting:
1267    for v in self.glyphs.itervalues():
1268      if hasattr(v, "data"):
1269        v.dropInstructionsFast()
1270      else:
1271        v.program = fontTools.ttLib.tables.ttProgram.Program()
1272        v.program.fromBytecode([])
1273  return True
1274
1275@_add_method(fontTools.ttLib.getTableClass('CFF '))
1276def prune_pre_subset(self, options):
1277  cff = self.cff
1278  # CFF table must have one font only
1279  cff.fontNames = cff.fontNames[:1]
1280
1281  if options.notdef_glyph and not options.notdef_outline:
1282    for fontname in cff.keys():
1283      font = cff[fontname]
1284      c,_ = font.CharStrings.getItemAndSelector('.notdef')
1285      # XXX we should preserve the glyph width
1286      c.bytecode = '\x0e' # endchar
1287      c.program = None
1288
1289  return True # bool(cff.fontNames)
1290
1291@_add_method(fontTools.ttLib.getTableClass('CFF '))
1292def subset_glyphs(self, s):
1293  cff = self.cff
1294  for fontname in cff.keys():
1295    font = cff[fontname]
1296    cs = font.CharStrings
1297
1298    # Load all glyphs
1299    for g in font.charset:
1300      if g not in s.glyphs: continue
1301      c,sel = cs.getItemAndSelector(g)
1302
1303    if cs.charStringsAreIndexed:
1304      indices = [i for i,g in enumerate(font.charset) if g in s.glyphs]
1305      csi = cs.charStringsIndex
1306      csi.items = [csi.items[i] for i in indices]
1307      csi.count = len(csi.items)
1308      del csi.file, csi.offsets
1309      if hasattr(font, "FDSelect"):
1310        sel = font.FDSelect
1311        sel.format = None
1312        sel.gidArray = [sel.gidArray[i] for i in indices]
1313      cs.charStrings = dict((g,indices.index(v))
1314                            for g,v in cs.charStrings.iteritems()
1315                            if g in s.glyphs)
1316    else:
1317      cs.charStrings = dict((g,v)
1318                            for g,v in cs.charStrings.iteritems()
1319                            if g in s.glyphs)
1320    font.charset = [g for g in font.charset if g in s.glyphs]
1321    font.numGlyphs = len(font.charset)
1322
1323  return True # any(cff[fontname].numGlyphs for fontname in cff.keys())
1324
1325@_add_method(fontTools.misc.psCharStrings.T2CharString)
1326def subset_subroutines(self, subrs, gsubrs):
1327  p = self.program
1328  assert len(p)
1329  for i in xrange(1, len(p)):
1330    if p[i] == 'callsubr':
1331      assert type(p[i-1]) is int
1332      p[i-1] = subrs._used.index(p[i-1] + subrs._old_bias) - subrs._new_bias
1333    elif p[i] == 'callgsubr':
1334      assert type(p[i-1]) is int
1335      p[i-1] = gsubrs._used.index(p[i-1] + gsubrs._old_bias) - gsubrs._new_bias
1336
1337@_add_method(fontTools.misc.psCharStrings.T2CharString)
1338def drop_hints(self):
1339  hints = self._hints
1340
1341  if hints.has_hint:
1342    self.program = self.program[hints.last_hint:]
1343    if hasattr(self, 'width'):
1344      # Insert width back if needed
1345      if self.width != self.private.defaultWidthX:
1346        self.program.insert(0, self.width - self.private.nominalWidthX)
1347
1348  if hints.has_hintmask:
1349    i = 0
1350    p = self.program
1351    while i < len(p):
1352      if p[i] in ['hintmask', 'cntrmask']:
1353        assert i + 1 <= len(p)
1354        del p[i:i+2]
1355        continue
1356      i += 1
1357
1358  assert len(self.program)
1359
1360  del self._hints
1361
1362class _MarkingT2Decompiler(fontTools.misc.psCharStrings.SimpleT2Decompiler):
1363
1364  def __init__(self, localSubrs, globalSubrs):
1365    fontTools.misc.psCharStrings.SimpleT2Decompiler.__init__(self,
1366                                                             localSubrs,
1367                                                             globalSubrs)
1368    for subrs in [localSubrs, globalSubrs]:
1369      if subrs and not hasattr(subrs, "_used"):
1370        subrs._used = set()
1371
1372  def op_callsubr(self, index):
1373    self.localSubrs._used.add(self.operandStack[-1]+self.localBias)
1374    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_callsubr(self, index)
1375
1376  def op_callgsubr(self, index):
1377    self.globalSubrs._used.add(self.operandStack[-1]+self.globalBias)
1378    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_callgsubr(self, index)
1379
1380class _DehintingT2Decompiler(fontTools.misc.psCharStrings.SimpleT2Decompiler):
1381
1382  class Hints:
1383    def __init__(self):
1384      # Whether calling this charstring produces any hint stems
1385      self.has_hint = False
1386      # Index to start at to drop all hints
1387      self.last_hint = 0
1388      # Index up to which we know more hints are possible.  Only
1389      # relevant if status is 0 or 1.
1390      self.last_checked = 0
1391      # The status means:
1392      # 0: after dropping hints, this charstring is empty
1393      # 1: after dropping hints, there may be more hints continuing after this
1394      # 2: no more hints possible after this charstring
1395      self.status = 0
1396      # Has hintmask instructions; not recursive
1397      self.has_hintmask = False
1398    pass
1399
1400  def __init__(self, css, localSubrs, globalSubrs):
1401    self._css = css
1402    fontTools.misc.psCharStrings.SimpleT2Decompiler.__init__(self,
1403                                                             localSubrs,
1404                                                             globalSubrs)
1405
1406  def execute(self, charString):
1407    old_hints = charString._hints if hasattr(charString, '_hints') else None
1408    charString._hints = self.Hints()
1409
1410    fontTools.misc.psCharStrings.SimpleT2Decompiler.execute(self, charString)
1411
1412    hints = charString._hints
1413
1414    if hints.has_hint or hints.has_hintmask:
1415      self._css.add(charString)
1416
1417    if hints.status != 2:
1418      # Check from last_check, make sure we didn't have any operators.
1419      for i in xrange(hints.last_checked, len(charString.program) - 1):
1420        if type(charString.program[i]) == str:
1421          hints.status = 2
1422          break;
1423        else:
1424          hints.status = 1 # There's *something* here
1425      hints.last_checked = len(charString.program)
1426
1427    if old_hints:
1428      assert hints.__dict__ == old_hints.__dict__
1429
1430  def op_callsubr(self, index):
1431    subr = self.localSubrs[self.operandStack[-1]+self.localBias]
1432    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_callsubr(self, index)
1433    self.processSubr(index, subr)
1434
1435  def op_callgsubr(self, index):
1436    subr = self.globalSubrs[self.operandStack[-1]+self.globalBias]
1437    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_callgsubr(self, index)
1438    self.processSubr(index, subr)
1439
1440  def op_hstem(self, index):
1441    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_hstem(self, index)
1442    self.processHint(index)
1443  def op_vstem(self, index):
1444    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_vstem(self, index)
1445    self.processHint(index)
1446  def op_hstemhm(self, index):
1447    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_hstemhm(self, index)
1448    self.processHint(index)
1449  def op_vstemhm(self, index):
1450    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_vstemhm(self, index)
1451    self.processHint(index)
1452  def op_hintmask(self, index):
1453    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_hintmask(self, index)
1454    self.processHintmask(index)
1455  def op_cntrmask(self, index):
1456    fontTools.misc.psCharStrings.SimpleT2Decompiler.op_cntrmask(self, index)
1457    self.processHintmask(index)
1458
1459  def processHintmask(self, index):
1460    cs = self.callingStack[-1]
1461    hints = cs._hints
1462    hints.has_hintmask = True
1463    if hints.status != 2 and hints.has_hint:
1464      # Check from last_check, see if we may be an implicit vstem
1465      for i in xrange(hints.last_checked, index - 1):
1466        if type(cs.program[i]) == str:
1467          hints.status = 2
1468          break;
1469      if hints.status != 2:
1470        # We are an implicit vstem
1471        hints.last_hint = index + 1
1472        hints.status = 0
1473      hints.last_checked = index + 1
1474
1475  def processHint(self, index):
1476    cs = self.callingStack[-1]
1477    hints = cs._hints
1478    hints.has_hint = True
1479    hints.last_hint = index
1480    hints.last_checked = index
1481
1482  def processSubr(self, index, subr):
1483    cs = self.callingStack[-1]
1484    hints = cs._hints
1485    subr_hints = subr._hints
1486
1487    if subr_hints.has_hint:
1488      if hints.status != 2:
1489        hints.has_hint = True
1490        hints.last_checked = index
1491        hints.status = subr_hints.status
1492        # Decide where to chop off from
1493        if subr_hints.status == 0:
1494          hints.last_hint = index
1495        else:
1496          hints.last_hint = index - 2 # Leave the subr call in
1497      else:
1498        # In my understanding, this is a font bug.  Ie. it has hint stems
1499        # *after* path construction.  I've seen this in widespread fonts.
1500        # Best to ignore the hints I suppose...
1501        pass
1502        #assert 0
1503    else:
1504      hints.status = max(hints.status, subr_hints.status)
1505      if hints.status != 2:
1506        # Check from last_check, make sure we didn't have
1507        # any operators.
1508        for i in xrange(hints.last_checked, index - 1):
1509          if type(cs.program[i]) == str:
1510            hints.status = 2
1511            break;
1512        hints.last_checked = index
1513
1514@_add_method(fontTools.ttLib.getTableClass('CFF '))
1515def prune_post_subset(self, options):
1516  cff = self.cff
1517  for fontname in cff.keys():
1518    font = cff[fontname]
1519    cs = font.CharStrings
1520
1521
1522    #
1523    # Drop unused FontDictionaries
1524    #
1525    if hasattr(font, "FDSelect"):
1526      sel = font.FDSelect
1527      indices = _uniq_sort(sel.gidArray)
1528      sel.gidArray = [indices.index (ss) for ss in sel.gidArray]
1529      arr = font.FDArray
1530      arr.items = [arr[i] for i in indices]
1531      arr.count = len(arr.items)
1532      del arr.file, arr.offsets
1533
1534
1535    #
1536    # Drop hints if not needed
1537    #
1538    if not options.hinting:
1539
1540      #
1541      # This can be tricky, but doesn't have to.  What we do is:
1542      #
1543      # - Run all used glyph charstrings and recurse into subroutines,
1544      # - For each charstring (including subroutines), if it has any
1545      #   of the hint stem operators, we mark it as such.  Upon returning,
1546      #   for each charstring we note all the subroutine calls it makes
1547      #   that (recursively) contain a stem,
1548      # - Dropping hinting then consists of the following two ops:
1549      #   * Drop the piece of the program in each charstring before the
1550      #     last call to a stem op or a stem-calling subroutine,
1551      #   * Drop all hintmask operations.
1552      # - It's trickier... A hintmask right after hints and a few numbers
1553      #   will act as an implicit vstemhm.  As such, we track whether
1554      #   we have seen any non-hint operators so far and do the right
1555      #   thing, recursively...  Good luck understanding that :(
1556      #
1557      css = set()
1558      for g in font.charset:
1559        c,sel = cs.getItemAndSelector(g)
1560        # Make sure it's decompiled.  We want our "decompiler" to walk
1561        # the program, not the bytecode.
1562        c.draw(fontTools.pens.basePen.NullPen())
1563        subrs = getattr(c.private, "Subrs", [])
1564        decompiler = _DehintingT2Decompiler(css, subrs, c.globalSubrs)
1565        decompiler.execute(c)
1566      for charstring in css:
1567        charstring.drop_hints()
1568
1569
1570    #
1571    # Renumber subroutines to remove unused ones
1572    #
1573
1574    # Mark all used subroutines
1575    for g in font.charset:
1576      c,sel = cs.getItemAndSelector(g)
1577      subrs = getattr(c.private, "Subrs", [])
1578      decompiler = _MarkingT2Decompiler(subrs, c.globalSubrs)
1579      decompiler.execute(c)
1580
1581    all_subrs = [font.GlobalSubrs]
1582    if hasattr(font, 'FDSelect'):
1583      all_subrs.extend(fd.Private.Subrs for fd in font.FDArray if hasattr(fd.Private, 'Subrs') and fd.Private.Subrs)
1584    elif hasattr(font.Private, 'Subrs') and font.Private.Subrs:
1585      all_subrs.append(font.Private.Subrs)
1586
1587    subrs = set(subrs) # Remove duplicates
1588
1589    # Prepare
1590    for subrs in all_subrs:
1591      if not hasattr(subrs, '_used'):
1592        subrs._used = set()
1593      subrs._used = _uniq_sort(subrs._used)
1594      subrs._old_bias = fontTools.misc.psCharStrings.calcSubrBias(subrs)
1595      subrs._new_bias = fontTools.misc.psCharStrings.calcSubrBias(subrs._used)
1596
1597    # Renumber glyph charstrings
1598    for g in font.charset:
1599      c,sel = cs.getItemAndSelector(g)
1600      subrs = getattr(c.private, "Subrs", [])
1601      c.subset_subroutines (subrs, font.GlobalSubrs)
1602
1603    # Renumber subroutines themselves
1604    for subrs in all_subrs:
1605
1606      if subrs == font.GlobalSubrs:
1607        if not hasattr(font, 'FDSelect') and hasattr(font.Private, 'Subrs'):
1608          local_subrs = font.Private.Subrs
1609        else:
1610          local_subrs = []
1611      else:
1612        local_subrs = subrs
1613
1614      subrs.items = [subrs.items[i] for i in subrs._used]
1615      subrs.count = len(subrs.items)
1616      del subrs.file
1617      if hasattr(subrs, 'offsets'):
1618        del subrs.offsets
1619
1620      for i in xrange (subrs.count):
1621        subrs[i].subset_subroutines (local_subrs, font.GlobalSubrs)
1622
1623    # Cleanup
1624    for subrs in all_subrs:
1625      del subrs._used, subrs._old_bias, subrs._new_bias
1626
1627  return True
1628
1629@_add_method(fontTools.ttLib.getTableClass('cmap'))
1630def closure_glyphs(self, s):
1631  tables = [t for t in self.tables
1632            if t.platformID == 3 and t.platEncID in [1, 10]]
1633  for u in s.unicodes_requested:
1634    found = False
1635    for table in tables:
1636      if u in table.cmap:
1637        s.glyphs.add(table.cmap[u])
1638        found = True
1639        break
1640    if not found:
1641      s.log("No glyph for Unicode value %s; skipping." % u)
1642
1643@_add_method(fontTools.ttLib.getTableClass('cmap'))
1644def prune_pre_subset(self, options):
1645  if not options.legacy_cmap:
1646    # Drop non-Unicode / non-Symbol cmaps
1647    self.tables = [t for t in self.tables
1648                   if t.platformID == 3 and t.platEncID in [0, 1, 10]]
1649  if not options.symbol_cmap:
1650    self.tables = [t for t in self.tables
1651                   if t.platformID == 3 and t.platEncID in [1, 10]]
1652  # TODO(behdad) Only keep one subtable?
1653  # For now, drop format=0 which can't be subset_glyphs easily?
1654  self.tables = [t for t in self.tables if t.format != 0]
1655  return bool(self.tables)
1656
1657@_add_method(fontTools.ttLib.getTableClass('cmap'))
1658def subset_glyphs(self, s):
1659  s.glyphs = s.glyphs_cmaped
1660  for t in self.tables:
1661    # For reasons I don't understand I need this here
1662    # to force decompilation of the cmap format 14.
1663    try:
1664      getattr(t, "asdf")
1665    except AttributeError:
1666      pass
1667    if t.format == 14:
1668      # TODO(behdad) XXX We drop all the default-UVS mappings(g==None).
1669      t.uvsDict = dict((v,[(u,g) for u,g in l if g in s.glyphs])
1670                       for v,l in t.uvsDict.iteritems())
1671      t.uvsDict = dict((v,l) for v,l in t.uvsDict.iteritems() if l)
1672    else:
1673      t.cmap = dict((u,g) for u,g in t.cmap.iteritems()
1674                    if g in s.glyphs_requested or u in s.unicodes_requested)
1675  self.tables = [t for t in self.tables
1676                 if (t.cmap if t.format != 14 else t.uvsDict)]
1677  # TODO(behdad) Convert formats when needed.
1678  # In particular, if we have a format=12 without non-BMP
1679  # characters, either drop format=12 one or convert it
1680  # to format=4 if there's not one.
1681  return bool(self.tables)
1682
1683@_add_method(fontTools.ttLib.getTableClass('name'))
1684def prune_pre_subset(self, options):
1685  if '*' not in options.name_IDs:
1686    self.names = [n for n in self.names if n.nameID in options.name_IDs]
1687  if not options.name_legacy:
1688    self.names = [n for n in self.names
1689                  if n.platformID == 3 and n.platEncID == 1]
1690  if '*' not in options.name_languages:
1691    self.names = [n for n in self.names if n.langID in options.name_languages]
1692  return True  # Retain even if empty
1693
1694
1695# TODO(behdad) OS/2 ulUnicodeRange / ulCodePageRange?
1696# TODO(behdad) Drop unneeded GSUB/GPOS Script/LangSys entries.
1697# TODO(behdad) Drop empty GSUB/GPOS, and GDEF if no GSUB/GPOS left
1698# TODO(behdad) Drop GDEF subitems if unused by lookups
1699# TODO(behdad) Avoid recursing too much (in GSUB/GPOS and in CFF)
1700# TODO(behdad) Text direction considerations.
1701# TODO(behdad) Text script / language considerations.
1702# TODO(behdad) Option to drop hmtx for CFF?
1703
1704
1705class Options(object):
1706
1707  class UnknownOptionError(Exception):
1708    pass
1709
1710  _drop_tables_default = ['BASE', 'JSTF', 'DSIG', 'EBDT', 'EBLC', 'EBSC', 'SVG ',
1711                          'PCLT', 'LTSH']
1712  _drop_tables_default += ['Feat', 'Glat', 'Gloc', 'Silf', 'Sill']  # Graphite
1713  _drop_tables_default += ['CBLC', 'CBDT', 'sbix', 'COLR', 'CPAL']  # Color
1714  _no_subset_tables_default = ['gasp', 'head', 'hhea', 'maxp', 'vhea', 'OS/2',
1715                               'loca', 'name', 'cvt ', 'fpgm', 'prep']
1716  _hinting_tables_default = ['cvt ', 'fpgm', 'prep', 'hdmx', 'VDMX']
1717
1718  # Based on HarfBuzz shapers
1719  _layout_features_groups = {
1720    # Default shaper
1721    'common': ['ccmp', 'liga', 'locl', 'mark', 'mkmk', 'rlig'],
1722    'horizontal': ['calt', 'clig', 'curs', 'kern', 'rclt'],
1723    'vertical':  ['valt', 'vert', 'vkrn', 'vpal', 'vrt2'],
1724    'ltr': ['ltra', 'ltrm'],
1725    'rtl': ['rtla', 'rtlm'],
1726    # Complex shapers
1727    'arabic': ['init', 'medi', 'fina', 'isol', 'med2', 'fin2', 'fin3',
1728               'cswh', 'mset'],
1729    'hangul': ['ljmo', 'vjmo', 'tjmo'],
1730    'tibetal': ['abvs', 'blws', 'abvm', 'blwm'],
1731    'indic': ['nukt', 'akhn', 'rphf', 'rkrf', 'pref', 'blwf', 'half',
1732              'abvf', 'pstf', 'cfar', 'vatu', 'cjct', 'init', 'pres',
1733              'abvs', 'blws', 'psts', 'haln', 'dist', 'abvm', 'blwm'],
1734  }
1735  _layout_features_default = _uniq_sort(sum(
1736      _layout_features_groups.itervalues(), []))
1737
1738  drop_tables = _drop_tables_default
1739  no_subset_tables = _no_subset_tables_default
1740  hinting_tables = _hinting_tables_default
1741  layout_features = _layout_features_default
1742  hinting = False
1743  glyph_names = False
1744  legacy_cmap = False
1745  symbol_cmap = False
1746  name_IDs = [1, 2]  # Family and Style
1747  name_legacy = False
1748  name_languages = [0x0409]  # English
1749  notdef_glyph = True # gid0 for TrueType / .notdef for CFF
1750  notdef_outline = False # No need for notdef to have an outline really
1751  recommended_glyphs = False  # gid1, gid2, gid3 for TrueType
1752  recalc_bounds = False # Recalculate font bounding boxes
1753  canonical_order = False # Order tables as recommended
1754  flavor = None # May be 'woff'
1755
1756  def __init__(self, **kwargs):
1757
1758    self.set(**kwargs)
1759
1760  def set(self, **kwargs):
1761    for k,v in kwargs.iteritems():
1762      if not hasattr(self, k):
1763        raise self.UnknownOptionError("Unknown option '%s'" % k)
1764      setattr(self, k, v)
1765
1766  def parse_opts(self, argv, ignore_unknown=False):
1767    ret = []
1768    opts = {}
1769    for a in argv:
1770      orig_a = a
1771      if not a.startswith('--'):
1772        ret.append(a)
1773        continue
1774      a = a[2:]
1775      i = a.find('=')
1776      op = '='
1777      if i == -1:
1778        if a.startswith("no-"):
1779          k = a[3:]
1780          v = False
1781        else:
1782          k = a
1783          v = True
1784      else:
1785        k = a[:i]
1786        if k[-1] in "-+":
1787          op = k[-1]+'='  # Ops is '-=' or '+=' now.
1788          k = k[:-1]
1789        v = a[i+1:]
1790      k = k.replace('-', '_')
1791      if not hasattr(self, k):
1792        if ignore_unknown == True or k in ignore_unknown:
1793          ret.append(orig_a)
1794          continue
1795        else:
1796          raise self.UnknownOptionError("Unknown option '%s'" % a)
1797
1798      ov = getattr(self, k)
1799      if isinstance(ov, bool):
1800        v = bool(v)
1801      elif isinstance(ov, int):
1802        v = int(v)
1803      elif isinstance(ov, list):
1804        vv = v.split(',')
1805        if vv == ['']:
1806          vv = []
1807        vv = [int(x, 0) if len(x) and x[0] in "0123456789" else x for x in vv]
1808        if op == '=':
1809          v = vv
1810        elif op == '+=':
1811          v = ov
1812          v.extend(vv)
1813        elif op == '-=':
1814          v = ov
1815          for x in vv:
1816            if x in v:
1817              v.remove(x)
1818        else:
1819          assert 0
1820
1821      opts[k] = v
1822    self.set(**opts)
1823
1824    return ret
1825
1826
1827class Subsetter(object):
1828
1829  def __init__(self, options=None, log=None):
1830
1831    if not log:
1832      log = Logger()
1833    if not options:
1834      options = Options()
1835
1836    self.options = options
1837    self.log = log
1838    self.unicodes_requested = set()
1839    self.glyphs_requested = set()
1840    self.glyphs = set()
1841
1842  def populate(self, glyphs=[], unicodes=[], text=""):
1843    self.unicodes_requested.update(unicodes)
1844    if isinstance(text, str):
1845      text = text.decode("utf8")
1846    for u in text:
1847      self.unicodes_requested.add(ord(u))
1848    self.glyphs_requested.update(glyphs)
1849    self.glyphs.update(glyphs)
1850
1851  def _prune_pre_subset(self, font):
1852
1853    for tag in font.keys():
1854      if tag == 'GlyphOrder': continue
1855
1856      if(tag in self.options.drop_tables or
1857         (tag in self.options.hinting_tables and not self.options.hinting)):
1858        self.log(tag, "dropped")
1859        del font[tag]
1860        continue
1861
1862      clazz = fontTools.ttLib.getTableClass(tag)
1863
1864      if hasattr(clazz, 'prune_pre_subset'):
1865        table = font[tag]
1866        self.log.lapse("load '%s'" % tag)
1867        retain = table.prune_pre_subset(self.options)
1868        self.log.lapse("prune  '%s'" % tag)
1869        if not retain:
1870          self.log(tag, "pruned to empty; dropped")
1871          del font[tag]
1872          continue
1873        else:
1874          self.log(tag, "pruned")
1875
1876  def _closure_glyphs(self, font):
1877
1878    self.glyphs = self.glyphs_requested.copy()
1879
1880    if 'cmap' in font:
1881      font['cmap'].closure_glyphs(self)
1882    self.glyphs_cmaped = self.glyphs
1883
1884    if self.options.notdef_glyph:
1885      if 'glyf' in font:
1886        self.glyphs.add(font.getGlyphName(0))
1887        self.log("Added gid0 to subset")
1888      else:
1889        self.glyphs.add('.notdef')
1890        self.log("Added .notdef to subset")
1891    if self.options.recommended_glyphs:
1892      if 'glyf' in font:
1893        for i in range(4):
1894          self.glyphs.add(font.getGlyphName(i))
1895        self.log("Added first four glyphs to subset")
1896
1897    if 'GSUB' in font:
1898      self.log("Closing glyph list over 'GSUB': %d glyphs before" %
1899                len(self.glyphs))
1900      self.log.glyphs(self.glyphs, font=font)
1901      font['GSUB'].closure_glyphs(self)
1902      self.log("Closed  glyph list over 'GSUB': %d glyphs after" %
1903                len(self.glyphs))
1904      self.log.glyphs(self.glyphs, font=font)
1905      self.log.lapse("close glyph list over 'GSUB'")
1906    self.glyphs_gsubed = self.glyphs.copy()
1907
1908    if 'glyf' in font:
1909      self.log("Closing glyph list over 'glyf': %d glyphs before" %
1910                len(self.glyphs))
1911      self.log.glyphs(self.glyphs, font=font)
1912      font['glyf'].closure_glyphs(self)
1913      self.log("Closed  glyph list over 'glyf': %d glyphs after" %
1914                len(self.glyphs))
1915      self.log.glyphs(self.glyphs, font=font)
1916      self.log.lapse("close glyph list over 'glyf'")
1917    self.glyphs_glyfed = self.glyphs.copy()
1918
1919    self.glyphs_all = self.glyphs.copy()
1920
1921    self.log("Retaining %d glyphs: " % len(self.glyphs_all))
1922
1923  def _subset_glyphs(self, font):
1924    for tag in font.keys():
1925      if tag == 'GlyphOrder': continue
1926      clazz = fontTools.ttLib.getTableClass(tag)
1927
1928      if tag in self.options.no_subset_tables:
1929        self.log(tag, "subsetting not needed")
1930      elif hasattr(clazz, 'subset_glyphs'):
1931        table = font[tag]
1932        self.glyphs = self.glyphs_all
1933        retain = table.subset_glyphs(self)
1934        self.glyphs = self.glyphs_all
1935        self.log.lapse("subset '%s'" % tag)
1936        if not retain:
1937          self.log(tag, "subsetted to empty; dropped")
1938          del font[tag]
1939        else:
1940          self.log(tag, "subsetted")
1941      else:
1942        self.log(tag, "NOT subset; don't know how to subset; dropped")
1943        del font[tag]
1944
1945    glyphOrder = font.getGlyphOrder()
1946    glyphOrder = [g for g in glyphOrder if g in self.glyphs_all]
1947    font.setGlyphOrder(glyphOrder)
1948    font._buildReverseGlyphOrderDict()
1949    self.log.lapse("subset GlyphOrder")
1950
1951  def _prune_post_subset(self, font):
1952    for tag in font.keys():
1953      if tag == 'GlyphOrder': continue
1954      clazz = fontTools.ttLib.getTableClass(tag)
1955      if hasattr(clazz, 'prune_post_subset'):
1956        table = font[tag]
1957        retain = table.prune_post_subset(self.options)
1958        self.log.lapse("prune  '%s'" % tag)
1959        if not retain:
1960          self.log(tag, "pruned to empty; dropped")
1961          del font[tag]
1962        else:
1963          self.log(tag, "pruned")
1964
1965  def subset(self, font):
1966
1967    self._prune_pre_subset(font)
1968    self._closure_glyphs(font)
1969    self._subset_glyphs(font)
1970    self._prune_post_subset(font)
1971
1972
1973class Logger(object):
1974
1975  def __init__(self, verbose=False, xml=False, timing=False):
1976    self.verbose = verbose
1977    self.xml = xml
1978    self.timing = timing
1979    self.last_time = self.start_time = time.time()
1980
1981  def parse_opts(self, argv):
1982    argv = argv[:]
1983    for v in ['verbose', 'xml', 'timing']:
1984      if "--"+v in argv:
1985        setattr(self, v, True)
1986        argv.remove("--"+v)
1987    return argv
1988
1989  def __call__(self, *things):
1990    if not self.verbose:
1991      return
1992    print ' '.join(str(x) for x in things)
1993
1994  def lapse(self, *things):
1995    if not self.timing:
1996      return
1997    new_time = time.time()
1998    print "Took %0.3fs to %s" %(new_time - self.last_time,
1999                                 ' '.join(str(x) for x in things))
2000    self.last_time = new_time
2001
2002  def glyphs(self, glyphs, font=None):
2003    self("Names: ", sorted(glyphs))
2004    if font:
2005      reverseGlyphMap = font.getReverseGlyphMap()
2006      self("Gids : ", sorted(reverseGlyphMap[g] for g in glyphs))
2007
2008  def font(self, font, file=sys.stdout):
2009    if not self.xml:
2010      return
2011    from fontTools.misc import xmlWriter
2012    writer = xmlWriter.XMLWriter(file)
2013    font.disassembleInstructions = False  # Work around ttLib bug
2014    for tag in font.keys():
2015      writer.begintag(tag)
2016      writer.newline()
2017      font[tag].toXML(writer, font)
2018      writer.endtag(tag)
2019      writer.newline()
2020
2021
2022def load_font(fontFile,
2023              options,
2024              checkChecksums=False,
2025              dontLoadGlyphNames=False):
2026
2027  font = fontTools.ttLib.TTFont(fontFile,
2028                                checkChecksums=checkChecksums,
2029                                recalcBBoxes=options.recalc_bounds)
2030
2031  # Hack:
2032  #
2033  # If we don't need glyph names, change 'post' class to not try to
2034  # load them.  It avoid lots of headache with broken fonts as well
2035  # as loading time.
2036  #
2037  # Ideally ttLib should provide a way to ask it to skip loading
2038  # glyph names.  But it currently doesn't provide such a thing.
2039  #
2040  if dontLoadGlyphNames:
2041    post = fontTools.ttLib.getTableClass('post')
2042    saved = post.decode_format_2_0
2043    post.decode_format_2_0 = post.decode_format_3_0
2044    f = font['post']
2045    if f.formatType == 2.0:
2046      f.formatType = 3.0
2047    post.decode_format_2_0 = saved
2048
2049  return font
2050
2051def save_font(font, outfile, options):
2052  if options.flavor and not hasattr(font, 'flavor'):
2053    raise Exception("fonttools version does not support flavors.")
2054  font.flavor = options.flavor
2055  font.save(outfile, reorderTables=options.canonical_order)
2056
2057def main(args):
2058
2059  log = Logger()
2060  args = log.parse_opts(args)
2061
2062  options = Options()
2063  args = options.parse_opts(args, ignore_unknown=['text'])
2064
2065  if len(args) < 2:
2066    print >>sys.stderr, "usage: pyftsubset font-file glyph... [--text=ABC]... [--option=value]..."
2067    sys.exit(1)
2068
2069  fontfile = args[0]
2070  args = args[1:]
2071
2072  dontLoadGlyphNames =(not options.glyph_names and
2073         all(any(g.startswith(p)
2074             for p in ['gid', 'glyph', 'uni', 'U+'])
2075              for g in args))
2076
2077  font = load_font(fontfile, options, dontLoadGlyphNames=dontLoadGlyphNames)
2078  subsetter = Subsetter(options=options, log=log)
2079  log.lapse("load font")
2080
2081  names = font.getGlyphNames()
2082  log.lapse("loading glyph names")
2083
2084  glyphs = []
2085  unicodes = []
2086  text = ""
2087  for g in args:
2088    if g == '*':
2089      glyphs.extend(font.getGlyphOrder())
2090      continue
2091    if g in names:
2092      glyphs.append(g)
2093      continue
2094    if g.startswith('--text='):
2095      text += g[7:]
2096      continue
2097    if g.startswith('uni') or g.startswith('U+'):
2098      if g.startswith('uni') and len(g) > 3:
2099        g = g[3:]
2100      elif g.startswith('U+') and len(g) > 2:
2101        g = g[2:]
2102      u = int(g, 16)
2103      unicodes.append(u)
2104      continue
2105    if g.startswith('gid') or g.startswith('glyph'):
2106      if g.startswith('gid') and len(g) > 3:
2107        g = g[3:]
2108      elif g.startswith('glyph') and len(g) > 5:
2109        g = g[5:]
2110      try:
2111        glyphs.append(font.getGlyphName(int(g), requireReal=1))
2112      except ValueError:
2113        raise Exception("Invalid glyph identifier: %s" % g)
2114      continue
2115    raise Exception("Invalid glyph identifier: %s" % g)
2116  log.lapse("compile glyph list")
2117  log("Unicodes:", unicodes)
2118  log("Glyphs:", glyphs)
2119
2120  subsetter.populate(glyphs=glyphs, unicodes=unicodes, text=text)
2121  subsetter.subset(font)
2122
2123  outfile = fontfile + '.subset'
2124
2125  save_font (font, outfile, options)
2126  log.lapse("compile and save font")
2127
2128  log.last_time = log.start_time
2129  log.lapse("make one with everything(TOTAL TIME)")
2130
2131  if log.verbose:
2132    import os
2133    log("Input  font: %d bytes" % os.path.getsize(fontfile))
2134    log("Subset font: %d bytes" % os.path.getsize(outfile))
2135
2136  log.font(font)
2137
2138  font.close()
2139
2140
2141__all__ = [
2142  'Options',
2143  'Subsetter',
2144  'Logger',
2145  'load_font',
2146  'save_font',
2147  'main'
2148]
2149
2150if __name__ == '__main__':
2151  main(sys.argv[1:])
2152