fontchain_lint.py revision 6e06ad055b35b197b3083728c6c5d311fb12e57a 
1#!/usr/bin/env python
2
3import collections
4import copy
5import glob
6import itertools
7from os import path
8import sys
9from xml.etree import ElementTree
10
11from fontTools import ttLib
12
13EMOJI_VS = 0xFE0F
14
15LANG_TO_SCRIPT = {
16    'as': 'Beng',
17    'bg': 'Cyrl',
18    'bn': 'Beng',
19    'cu': 'Cyrl',
20    'cy': 'Latn',
21    'da': 'Latn',
22    'de': 'Latn',
23    'en': 'Latn',
24    'es': 'Latn',
25    'et': 'Latn',
26    'eu': 'Latn',
27    'fr': 'Latn',
28    'ga': 'Latn',
29    'gu': 'Gujr',
30    'hi': 'Deva',
31    'hr': 'Latn',
32    'hu': 'Latn',
33    'hy': 'Armn',
34    'ja': 'Jpan',
35    'kn': 'Knda',
36    'ko': 'Kore',
37    'ml': 'Mlym',
38    'mn': 'Cyrl',
39    'mr': 'Deva',
40    'nb': 'Latn',
41    'nn': 'Latn',
42    'or': 'Orya',
43    'pa': 'Guru',
44    'pt': 'Latn',
45    'sl': 'Latn',
46    'ta': 'Taml',
47    'te': 'Telu',
48    'tk': 'Latn',
49}
50
51def lang_to_script(lang_code):
52    lang = lang_code.lower()
53    while lang not in LANG_TO_SCRIPT:
54        hyphen_idx = lang.rfind('-')
55        assert hyphen_idx != -1, (
56            'We do not know what script the "%s" language is written in.'
57            % lang_code)
58        assumed_script = lang[hyphen_idx+1:]
59        if len(assumed_script) == 4 and assumed_script.isalpha():
60            # This is actually the script
61            return assumed_script.title()
62        lang = lang[:hyphen_idx]
63    return LANG_TO_SCRIPT[lang]
64
65
66def printable(inp):
67    if type(inp) is set:  # set of character sequences
68        return '{' + ', '.join([printable(seq) for seq in inp]) + '}'
69    if type(inp) is tuple:  # character sequence
70        return '<' + (', '.join([printable(ch) for ch in inp])) + '>'
71    else:  # single character
72        return 'U+%04X' % inp
73
74
75def open_font(font):
76    font_file, index = font
77    font_path = path.join(_fonts_dir, font_file)
78    if index is not None:
79        return ttLib.TTFont(font_path, fontNumber=index)
80    else:
81        return ttLib.TTFont(font_path)
82
83
84def get_best_cmap(font):
85    ttfont = open_font(font)
86    all_unicode_cmap = None
87    bmp_cmap = None
88    for cmap in ttfont['cmap'].tables:
89        specifier = (cmap.format, cmap.platformID, cmap.platEncID)
90        if specifier == (4, 3, 1):
91            assert bmp_cmap is None, 'More than one BMP cmap in %s' % (font, )
92            bmp_cmap = cmap
93        elif specifier == (12, 3, 10):
94            assert all_unicode_cmap is None, (
95                'More than one UCS-4 cmap in %s' % (font, ))
96            all_unicode_cmap = cmap
97
98    return all_unicode_cmap.cmap if all_unicode_cmap else bmp_cmap.cmap
99
100
101def get_variation_sequences_cmap(font):
102    ttfont = open_font(font)
103    vs_cmap = None
104    for cmap in ttfont['cmap'].tables:
105        specifier = (cmap.format, cmap.platformID, cmap.platEncID)
106        if specifier == (14, 0, 5):
107            assert vs_cmap is None, 'More than one VS cmap in %s' % (font, )
108            vs_cmap = cmap
109    return vs_cmap
110
111
112def get_emoji_map(font):
113    # Add normal characters
114    emoji_map = copy.copy(get_best_cmap(font))
115    reverse_cmap = {glyph: code for code, glyph in emoji_map.items()}
116
117    # Add variation sequences
118    vs_dict = get_variation_sequences_cmap(font).uvsDict
119    for vs in vs_dict:
120        for base, glyph in vs_dict[vs]:
121            if glyph is None:
122                emoji_map[(base, vs)] = emoji_map[base]
123            else:
124                emoji_map[(base, vs)] = glyph
125
126    # Add GSUB rules
127    ttfont = open_font(font)
128    for lookup in ttfont['GSUB'].table.LookupList.Lookup:
129        if lookup.LookupType != 4:
130            # Other lookups are used in the emoji font for fallback.
131            # We ignore them for now.
132            continue
133        for subtable in lookup.SubTable:
134            ligatures = subtable.ligatures
135            for first_glyph in ligatures:
136                for ligature in ligatures[first_glyph]:
137                    sequence = [first_glyph] + ligature.Component
138                    sequence = [reverse_cmap[glyph] for glyph in sequence]
139                    sequence = tuple(sequence)
140                    # Make sure no starting subsequence of 'sequence' has been
141                    # seen before.
142                    for sub_len in range(2, len(sequence)+1):
143                        subsequence = sequence[:sub_len]
144                        assert subsequence not in emoji_map
145                    emoji_map[sequence] = ligature.LigGlyph
146
147    return emoji_map
148
149
150def assert_font_supports_any_of_chars(font, chars):
151    best_cmap = get_best_cmap(font)
152    for char in chars:
153        if char in best_cmap:
154            return
155    sys.exit('None of characters in %s were found in %s' % (chars, font))
156
157
158def assert_font_supports_all_of_chars(font, chars):
159    best_cmap = get_best_cmap(font)
160    for char in chars:
161        assert char in best_cmap, (
162            'U+%04X was not found in %s' % (char, font))
163
164
165def assert_font_supports_none_of_chars(font, chars):
166    best_cmap = get_best_cmap(font)
167    for char in chars:
168        assert char not in best_cmap, (
169            'U+%04X was found in %s' % (char, font))
170
171
172def assert_font_supports_all_sequences(font, sequences):
173    vs_dict = get_variation_sequences_cmap(font).uvsDict
174    for base, vs in sorted(sequences):
175        assert vs in vs_dict and (base, None) in vs_dict[vs], (
176            '<U+%04X, U+%04X> was not found in %s' % (base, vs, font))
177
178
179def check_hyphens(hyphens_dir):
180    # Find all the scripts that need automatic hyphenation
181    scripts = set()
182    for hyb_file in glob.iglob(path.join(hyphens_dir, '*.hyb')):
183        hyb_file = path.basename(hyb_file)
184        assert hyb_file.startswith('hyph-'), (
185            'Unknown hyphenation file %s' % hyb_file)
186        lang_code = hyb_file[hyb_file.index('-')+1:hyb_file.index('.')]
187        scripts.add(lang_to_script(lang_code))
188
189    HYPHENS = {0x002D, 0x2010}
190    for script in scripts:
191        fonts = _script_to_font_map[script]
192        assert fonts, 'No fonts found for the "%s" script' % script
193        for font in fonts:
194            assert_font_supports_any_of_chars(font, HYPHENS)
195
196
197class FontRecord(object):
198    def __init__(self, name, scripts, variant, weight, style, font):
199        self.name = name
200        self.scripts = scripts
201        self.variant = variant
202        self.weight = weight
203        self.style = style
204        self.font = font
205
206
207def parse_fonts_xml(fonts_xml_path):
208    global _script_to_font_map, _fallback_chain
209    _script_to_font_map = collections.defaultdict(set)
210    _fallback_chain = []
211    tree = ElementTree.parse(fonts_xml_path)
212    families = tree.findall('family')
213    # Minikin supports up to 254 but users can place their own font at the first
214    # place. Thus, 253 is the maximum allowed number of font families in the
215    # default collection.
216    assert len(families) < 254, (
217        'System font collection can contains up to 253 font families.')
218    for family in families:
219        name = family.get('name')
220        variant = family.get('variant')
221        langs = family.get('lang')
222        if name:
223            assert variant is None, (
224                'No variant expected for LGC font %s.' % name)
225            assert langs is None, (
226                'No language expected for LGC fonts %s.' % name)
227        else:
228            assert variant in {None, 'elegant', 'compact'}, (
229                'Unexpected value for variant: %s' % variant)
230
231        if langs:
232            langs = langs.split()
233            scripts = {lang_to_script(lang) for lang in langs}
234        else:
235            scripts = set()
236
237        for child in family:
238            assert child.tag == 'font', (
239                'Unknown tag <%s>' % child.tag)
240            font_file = child.text.rstrip()
241            weight = int(child.get('weight'))
242            assert weight % 100 == 0, (
243                'Font weight "%d" is not a multiple of 100.' % weight)
244
245            style = child.get('style')
246            assert style in {'normal', 'italic'}, (
247                'Unknown style "%s"' % style)
248
249            index = child.get('index')
250            if index:
251                index = int(index)
252
253            _fallback_chain.append(FontRecord(
254                name,
255                frozenset(scripts),
256                variant,
257                weight,
258                style,
259                (font_file, index)))
260
261            if name: # non-empty names are used for default LGC fonts
262                map_scripts = {'Latn', 'Grek', 'Cyrl'}
263            else:
264                map_scripts = scripts
265            for script in map_scripts:
266                _script_to_font_map[script].add((font_file, index))
267
268
269def check_emoji_coverage(all_emoji, equivalent_emoji):
270    emoji_font = get_emoji_font()
271    check_emoji_font_coverage(emoji_font, all_emoji, equivalent_emoji)
272
273
274def get_emoji_font():
275    emoji_fonts = [
276        record.font for record in _fallback_chain
277        if 'Zsye' in record.scripts]
278    assert len(emoji_fonts) == 1, 'There are %d emoji fonts.' % len(emoji_fonts)
279    return emoji_fonts[0]
280
281
282def check_emoji_font_coverage(emoji_font, all_emoji, equivalent_emoji):
283    coverage = get_emoji_map(emoji_font)
284    for sequence in all_emoji:
285        assert sequence in coverage, (
286            '%s is not supported in the emoji font.' % printable(sequence))
287
288    for sequence in coverage:
289        if sequence in {0x0000, 0x000D, 0x0020}:
290            # The font needs to support a few extra characters, which is OK
291            continue
292        assert sequence in all_emoji, (
293            'Emoji font should not support %s.' % printable(sequence))
294
295    for first, second in sorted(equivalent_emoji.items()):
296        assert coverage[first] == coverage[second], (
297            '%s and %s should map to the same glyph.' % (
298                printable(first),
299                printable(second)))
300
301    for glyph in set(coverage.values()):
302        maps_to_glyph = [seq for seq in coverage if coverage[seq] == glyph]
303        if len(maps_to_glyph) > 1:
304            # There are more than one sequences mapping to the same glyph. We
305            # need to make sure they were expected to be equivalent.
306            equivalent_seqs = set()
307            for seq in maps_to_glyph:
308                equivalent_seq = seq
309                while equivalent_seq in equivalent_emoji:
310                    equivalent_seq = equivalent_emoji[equivalent_seq]
311                equivalent_seqs.add(equivalent_seq)
312            assert len(equivalent_seqs) == 1, (
313                'The sequences %s should not result in the same glyph %s' % (
314                    printable(equivalent_seqs),
315                    glyph))
316
317
318def check_emoji_defaults(default_emoji):
319    missing_text_chars = _emoji_properties['Emoji'] - default_emoji
320    emoji_font_seen = False
321    for record in _fallback_chain:
322        if 'Zsye' in record.scripts:
323            emoji_font_seen = True
324            # No need to check the emoji font
325            continue
326        # For later fonts, we only check them if they have a script
327        # defined, since the defined script may get them to a higher
328        # score even if they appear after the emoji font. However,
329        # we should skip checking the text symbols font, since
330        # symbol fonts should be able to override the emoji display
331        # style when 'Zsym' is explicitly specified by the user.
332        if emoji_font_seen and (not record.scripts or 'Zsym' in record.scripts):
333            continue
334
335        # Check default emoji-style characters
336        assert_font_supports_none_of_chars(record.font, sorted(default_emoji))
337
338        # Mark default text-style characters appearing in fonts above the emoji
339        # font as seen
340        if not emoji_font_seen:
341            missing_text_chars -= set(get_best_cmap(record.font))
342
343    # Noto does not have monochrome glyphs for Unicode 7.0 wingdings and
344    # webdings yet.
345    missing_text_chars -= _chars_by_age['7.0']
346    assert missing_text_chars == set(), (
347        'Text style version of some emoji characters are missing: ' +
348            repr(missing_text_chars))
349
350
351# Setting reverse to true returns a dictionary that maps the values to sets of
352# characters, useful for some binary properties. Otherwise, we get a
353# dictionary that maps characters to the property values, assuming there's only
354# one property in the file.
355def parse_unicode_datafile(file_path, reverse=False):
356    if reverse:
357        output_dict = collections.defaultdict(set)
358    else:
359        output_dict = {}
360    with open(file_path) as datafile:
361        for line in datafile:
362            if '#' in line:
363                line = line[:line.index('#')]
364            line = line.strip()
365            if not line:
366                continue
367
368            chars, prop = line.split(';')[:2]
369            chars = chars.strip()
370            prop = prop.strip()
371
372            if ' ' in chars:  # character sequence
373                sequence = [int(ch, 16) for ch in chars.split(' ')]
374                additions = [tuple(sequence)]
375            elif '..' in chars:  # character range
376                char_start, char_end = chars.split('..')
377                char_start = int(char_start, 16)
378                char_end = int(char_end, 16)
379                additions = xrange(char_start, char_end+1)
380            else:  # singe character
381                additions = [int(chars, 16)]
382            if reverse:
383                output_dict[prop].update(additions)
384            else:
385                for addition in additions:
386                    assert addition not in output_dict
387                    output_dict[addition] = prop
388    return output_dict
389
390
391def parse_emoji_variants(file_path):
392    emoji_set = set()
393    text_set = set()
394    with open(file_path) as datafile:
395        for line in datafile:
396            if '#' in line:
397                line = line[:line.index('#')]
398            line = line.strip()
399            if not line:
400                continue
401            sequence, description, _ = line.split(';')
402            sequence = sequence.strip().split(' ')
403            base = int(sequence[0], 16)
404            vs = int(sequence[1], 16)
405            description = description.strip()
406            if description == 'text style':
407                text_set.add((base, vs))
408            elif description == 'emoji style':
409                emoji_set.add((base, vs))
410    return text_set, emoji_set
411
412
413def parse_ucd(ucd_path):
414    global _emoji_properties, _chars_by_age
415    global _text_variation_sequences, _emoji_variation_sequences
416    global _emoji_sequences, _emoji_zwj_sequences
417    _emoji_properties = parse_unicode_datafile(
418        path.join(ucd_path, 'emoji-data.txt'), reverse=True)
419    emoji_properties_additions = parse_unicode_datafile(
420        path.join(ucd_path, 'additions', 'emoji-data.txt'), reverse=True)
421    for prop in emoji_properties_additions.keys():
422        _emoji_properties[prop].update(emoji_properties_additions[prop])
423
424    _chars_by_age = parse_unicode_datafile(
425        path.join(ucd_path, 'DerivedAge.txt'), reverse=True)
426    sequences = parse_emoji_variants(
427        path.join(ucd_path, 'emoji-variation-sequences.txt'))
428    _text_variation_sequences, _emoji_variation_sequences = sequences
429    _emoji_sequences = parse_unicode_datafile(
430        path.join(ucd_path, 'emoji-sequences.txt'))
431    _emoji_sequences.update(parse_unicode_datafile(
432        path.join(ucd_path, 'additions', 'emoji-sequences.txt')))
433    _emoji_zwj_sequences = parse_unicode_datafile(
434        path.join(ucd_path, 'emoji-zwj-sequences.txt'))
435    _emoji_zwj_sequences.update(parse_unicode_datafile(
436        path.join(ucd_path, 'additions', 'emoji-zwj-sequences.txt')))
437
438
439def flag_sequence(territory_code):
440    return tuple(0x1F1E6 + ord(ch) - ord('A') for ch in territory_code)
441
442
443UNSUPPORTED_FLAGS = frozenset({
444    flag_sequence('BL'), flag_sequence('BQ'), flag_sequence('DG'),
445    flag_sequence('EA'), flag_sequence('EH'), flag_sequence('FK'),
446    flag_sequence('GF'), flag_sequence('GP'), flag_sequence('GS'),
447    flag_sequence('MF'), flag_sequence('MQ'), flag_sequence('NC'),
448    flag_sequence('PM'), flag_sequence('RE'), flag_sequence('TF'),
449    flag_sequence('WF'), flag_sequence('XK'), flag_sequence('YT'),
450})
451
452EQUIVALENT_FLAGS = {
453    flag_sequence('BV'): flag_sequence('NO'),
454    flag_sequence('CP'): flag_sequence('FR'),
455    flag_sequence('HM'): flag_sequence('AU'),
456    flag_sequence('SJ'): flag_sequence('NO'),
457    flag_sequence('UM'): flag_sequence('US'),
458}
459
460COMBINING_KEYCAP = 0x20E3
461
462LEGACY_ANDROID_EMOJI = {
463    0xFE4E5: flag_sequence('JP'),
464    0xFE4E6: flag_sequence('US'),
465    0xFE4E7: flag_sequence('FR'),
466    0xFE4E8: flag_sequence('DE'),
467    0xFE4E9: flag_sequence('IT'),
468    0xFE4EA: flag_sequence('GB'),
469    0xFE4EB: flag_sequence('ES'),
470    0xFE4EC: flag_sequence('RU'),
471    0xFE4ED: flag_sequence('CN'),
472    0xFE4EE: flag_sequence('KR'),
473    0xFE82C: (ord('#'), COMBINING_KEYCAP),
474    0xFE82E: (ord('1'), COMBINING_KEYCAP),
475    0xFE82F: (ord('2'), COMBINING_KEYCAP),
476    0xFE830: (ord('3'), COMBINING_KEYCAP),
477    0xFE831: (ord('4'), COMBINING_KEYCAP),
478    0xFE832: (ord('5'), COMBINING_KEYCAP),
479    0xFE833: (ord('6'), COMBINING_KEYCAP),
480    0xFE834: (ord('7'), COMBINING_KEYCAP),
481    0xFE835: (ord('8'), COMBINING_KEYCAP),
482    0xFE836: (ord('9'), COMBINING_KEYCAP),
483    0xFE837: (ord('0'), COMBINING_KEYCAP),
484}
485
486ZWJ_IDENTICALS = {
487    # KISS
488    (0x1F469, 0x200D, 0x2764, 0x200D, 0x1F48B, 0x200D, 0x1F468): 0x1F48F,
489    # COUPLE WITH HEART
490    (0x1F469, 0x200D, 0x2764, 0x200D, 0x1F468): 0x1F491,
491    # FAMILY
492    (0x1F468, 0x200D, 0x1F469, 0x200D, 0x1F466): 0x1F46A,
493}
494
495ZWJ = 0x200D
496FEMALE_SIGN = 0x2640
497MALE_SIGN = 0x2642
498
499GENDER_DEFAULTS = [
500    (0x26F9, MALE_SIGN), # PERSON WITH BALL
501    (0x1F3C3, MALE_SIGN), # RUNNER
502    (0x1F3C4, MALE_SIGN), # SURFER
503    (0x1F3CA, MALE_SIGN), # SWIMMER
504    (0x1F3CB, MALE_SIGN), # WEIGHT LIFTER
505    (0x1F3CC, MALE_SIGN), # GOLFER
506    (0x1F46E, MALE_SIGN), # POLICE OFFICER
507    (0x1F46F, FEMALE_SIGN), # WOMAN WITH BUNNY EARS
508    (0x1F471, MALE_SIGN), # PERSON WITH BLOND HAIR
509    (0x1F473, MALE_SIGN), # MAN WITH TURBAN
510    (0x1F477, MALE_SIGN), # CONSTRUCTION WORKER
511    (0x1F481, FEMALE_SIGN), # INFORMATION DESK PERSON
512    (0x1F482, MALE_SIGN), # GUARDSMAN
513    (0x1F486, FEMALE_SIGN), # FACE MASSAGE
514    (0x1F487, FEMALE_SIGN), # HAIRCUT
515    (0x1F575, MALE_SIGN), # SLEUTH OR SPY
516    (0x1F645, FEMALE_SIGN), # FACE WITH NO GOOD GESTURE
517    (0x1F646, FEMALE_SIGN), # FACE WITH OK GESTURE
518    (0x1F647, MALE_SIGN), # PERSON BOWING DEEPLY
519    (0x1F64B, FEMALE_SIGN), # HAPPY PERSON RAISING ONE HAND
520    (0x1F64D, FEMALE_SIGN), # PERSON FROWNING
521    (0x1F64E, FEMALE_SIGN), # PERSON WITH POUTING FACE
522    (0x1F6A3, MALE_SIGN), # ROWBOAT
523    (0x1F6B4, MALE_SIGN), # BICYCLIST
524    (0x1F6B5, MALE_SIGN), # MOUNTAIN BICYCLIST
525    (0x1F6B6, MALE_SIGN), # PEDESTRIAN
526    (0x1F926, FEMALE_SIGN), # FACE PALM
527    (0x1F937, FEMALE_SIGN), # SHRUG
528    (0x1F938, MALE_SIGN), # PERSON DOING CARTWHEEL
529    (0x1F939, MALE_SIGN), # JUGGLING
530    (0x1F93C, MALE_SIGN), # WRESTLERS
531    (0x1F93D, MALE_SIGN), # WATER POLO
532    (0x1F93E, MALE_SIGN), # HANDBALL
533    (0x1F9D6, FEMALE_SIGN), # PERSON IN STEAMY ROOM
534    (0x1F9D7, FEMALE_SIGN), # PERSON CLIMBING
535    (0x1F9D8, FEMALE_SIGN), # PERSON IN LOTUS POSITION
536    (0x1F9D9, FEMALE_SIGN), # MAGE
537    (0x1F9DA, FEMALE_SIGN), # FAIRY
538    (0x1F9DB, FEMALE_SIGN), # VAMPIRE
539    (0x1F9DC, FEMALE_SIGN), # MERPERSON
540    (0x1F9DD, FEMALE_SIGN), # ELF
541    (0x1F9DE, FEMALE_SIGN), # GENIE
542    (0x1F9DF, FEMALE_SIGN), # ZOMBIE
543]
544
545def is_fitzpatrick_modifier(cp):
546    return 0x1F3FB <= cp <= 0x1F3FF
547
548
549def reverse_emoji(seq):
550    rev = list(reversed(seq))
551    # if there are fitzpatrick modifiers in the sequence, keep them after
552    # the emoji they modify
553    for i in xrange(1, len(rev)):
554        if is_fitzpatrick_modifier(rev[i-1]):
555            rev[i], rev[i-1] = rev[i-1], rev[i]
556    return tuple(rev)
557
558
559def compute_expected_emoji():
560    equivalent_emoji = {}
561    sequence_pieces = set()
562    all_sequences = set()
563    all_sequences.update(_emoji_variation_sequences)
564
565    # add zwj sequences not in the current emoji-zwj-sequences.txt
566    adjusted_emoji_zwj_sequences = dict(_emoji_zwj_sequences)
567    adjusted_emoji_zwj_sequences.update(_emoji_zwj_sequences)
568
569    # Add empty flag tag sequence that is supported as fallback
570    _emoji_sequences[(0x1F3F4, 0xE007F)] = 'Emoji_Tag_Sequence'
571
572    for sequence in _emoji_sequences.keys():
573        sequence = tuple(ch for ch in sequence if ch != EMOJI_VS)
574        all_sequences.add(sequence)
575        sequence_pieces.update(sequence)
576        if _emoji_sequences.get(sequence, None) == 'Emoji_Tag_Sequence':
577            # Add reverse of all emoji ZWJ sequences, which are added to the fonts
578            # as a workaround to get the sequences work in RTL text.
579            # TODO: test if these are actually needed by Minikin/HarfBuzz.
580            reversed_seq = reverse_emoji(sequence)
581            all_sequences.add(reversed_seq)
582            equivalent_emoji[reversed_seq] = sequence
583
584    for sequence in adjusted_emoji_zwj_sequences.keys():
585        sequence = tuple(ch for ch in sequence if ch != EMOJI_VS)
586        all_sequences.add(sequence)
587        sequence_pieces.update(sequence)
588        # Add reverse of all emoji ZWJ sequences, which are added to the fonts
589        # as a workaround to get the sequences work in RTL text.
590        reversed_seq = reverse_emoji(sequence)
591        all_sequences.add(reversed_seq)
592        equivalent_emoji[reversed_seq] = sequence
593
594    # Remove unsupported flags
595    all_sequences.difference_update(UNSUPPORTED_FLAGS)
596
597    # Add all tag characters used in flags
598    sequence_pieces.update(range(0xE0030, 0xE0039 + 1))
599    sequence_pieces.update(range(0xE0061, 0xE007A + 1))
600
601    all_emoji = (
602        _emoji_properties['Emoji'] |
603        all_sequences |
604        sequence_pieces |
605        set(LEGACY_ANDROID_EMOJI.keys()))
606    default_emoji = (
607        _emoji_properties['Emoji_Presentation'] |
608        all_sequences |
609        set(LEGACY_ANDROID_EMOJI.keys()))
610
611    equivalent_emoji.update(EQUIVALENT_FLAGS)
612    equivalent_emoji.update(LEGACY_ANDROID_EMOJI)
613    equivalent_emoji.update(ZWJ_IDENTICALS)
614
615    for ch, gender in GENDER_DEFAULTS:
616        equivalent_emoji[(ch, ZWJ, gender)] = ch
617        for skin_tone in range(0x1F3FB, 0x1F3FF+1):
618            skin_toned = (ch, skin_tone, ZWJ, gender)
619            if skin_toned in all_emoji:
620                equivalent_emoji[skin_toned] = (ch, skin_tone)
621
622    for seq in _emoji_variation_sequences:
623        equivalent_emoji[seq] = seq[0]
624
625    return all_emoji, default_emoji, equivalent_emoji
626
627
628def check_vertical_metrics():
629    for record in _fallback_chain:
630        if record.name in ['sans-serif', 'sans-serif-condensed']:
631            font = open_font(record.font)
632            assert font['head'].yMax == 2163 and font['head'].yMin == -555, (
633                'yMax and yMin of %s do not match expected values.' % (record.font,))
634
635        if record.name in ['sans-serif', 'sans-serif-condensed', 'serif', 'monospace']:
636            font = open_font(record.font)
637            assert font['hhea'].ascent == 1900 and font['hhea'].descent == -500, (
638                'ascent and descent of %s do not match expected values.' % (record.font,))
639
640
641def main():
642    global _fonts_dir
643    target_out = sys.argv[1]
644    _fonts_dir = path.join(target_out, 'fonts')
645
646    fonts_xml_path = path.join(target_out, 'etc', 'fonts.xml')
647    parse_fonts_xml(fonts_xml_path)
648
649    check_vertical_metrics()
650
651    hyphens_dir = path.join(target_out, 'usr', 'hyphen-data')
652    check_hyphens(hyphens_dir)
653
654    check_emoji = sys.argv[2]
655    if check_emoji == 'true':
656        ucd_path = sys.argv[3]
657        parse_ucd(ucd_path)
658        all_emoji, default_emoji, equivalent_emoji = compute_expected_emoji()
659        check_emoji_coverage(all_emoji, equivalent_emoji)
660        check_emoji_defaults(default_emoji)
661
662
663if __name__ == '__main__':
664    main()
665