Expand scripts for webfont generation.

6 files changed, 176 insertions, 141 deletions

diff --git a/Makefile b/Makefile
index f6da6a0..05196fa 100644
--- a/Makefile
+++ b/Makefile
@@ -58,12 +58,11 @@ web:
 	mkdir -p out/web
 	for source in hinted/*.ttf; do \
 		touched=$$(mktemp); \
-		subsetted=$$(mktemp); \
 		final=out/web/$$(basename $$source); \
+		final="$${final/Roboto/RobotoDraft}"; \
 		python scripts/touchup_for_web.py $$source $$touched && \
-		python scripts/subset_for_web.py $$touched $$subsetted && \
-		python scripts/force_yminmax.py $$subsetted $$final && \
-		rm $$touched $$subsetted; \
+		python scripts/subset_for_web.py $$touched $$final && \
+		rm $$touched; \
 	done
 
 test: test-android test-coverage test-general
diff --git a/res/buildnumber.txt b/res/buildnumber.txt
index 1a11189..210f3c5 100644
--- a/res/buildnumber.txt
+++ b/res/buildnumber.txt
@@ -1 +1 @@
-01047
\ No newline at end of file
+01171
\ No newline at end of file
diff --git a/res/web_charset.txt b/res/charsets/web.txt
index ee4d0c8..ee4d0c8 100644
--- a/res/web_charset.txt
+++ b/res/charsets/web.txt
diff --git a/scripts/run_web_tests.py b/scripts/run_web_tests.py
index 779abdc..fdf0fe6 100755
--- a/scripts/run_web_tests.py
+++ b/scripts/run_web_tests.py
@@ -1,12 +1,15 @@
 #!/usr/bin/python
-"""Test assumptions that Android relies on."""
+"""Test assumptions that web fonts rely on."""
 
 import glob
+import json
 import unittest
 
 from fontTools import ttLib
 from nototools import coverage
 from nototools import font_data
+from nototools import render
+from nototools import unicode_data
 
 
 def load_fonts():
@@ -111,6 +114,45 @@ class TestDigitWidths(unittest.TestCase):
             self.assertEqual(len(set(widths)), 1)
 
 
+class TestNames(unittest.TestCase):
+    """Tests various strings in the name table."""
+
+    def setUp(self):
+        self.family_name = 'RobotoDraft'
+        _, self.fonts = load_fonts()
+        self.names = []
+        for font in self.fonts:
+            self.names.append(font_data.get_name_records(font))
+
+    def test_copyright(self):
+        """Tests the copyright message."""
+        for records in self.names:
+            self.assertEqual(
+                records[0],
+                'Copyright 2014 Google Inc. All Rights Reserved.')
+
+    def test_family_name(self):
+        """Tests the family name."""
+        for records in self.names:
+            self.assertEqual(records[1], self.family_name)
+            if 16 in records:
+                self.assertEqual(records[16], self.family_name)
+
+    def test_unique_identifier_and_full_name(self):
+        """Tests the unique identifier and full name."""
+        for records in self.names:
+            expected_name = records[1] + ' ' + records[2]
+            self.assertEqual(records[3], expected_name)
+            self.assertEqual(records[4], expected_name)
+            self.assertFalse(records.has_key(18))
+
+    def test_postscript_name(self):
+        """Tests the postscript name."""
+        for records in self.names:
+            expected_name = records[1] + '-' + records[2].replace(' ', '')
+            self.assertEqual(records[6], expected_name)
+
+
 class TestHints(unittest.TestCase):
     """Tests hints."""
 
@@ -121,9 +163,9 @@ class TestHints(unittest.TestCase):
         bad_glyphs = ['uniFB01', 'uniFB02', 'uniFB03',
                       'uniFB04', 'uniFFFC', 'uni048C']
         self.known_missing_hints = [
-            (g, 'Roboto Light Italic') for g in bad_glyphs]
+            (g, 'RobotoDraft Light Italic') for g in bad_glyphs]
 
-    def test_digit_widths(self):
+    def test_existance_of_hints(self):
         """Tests all glyphs and makes sure non-composite ones have hints."""
         missing_hints = []
         for font in self.fonts:
@@ -139,5 +181,53 @@ class TestHints(unittest.TestCase):
         self.assertTrue(missing_hints <= self.known_missing_hints)
 
 
+class TestSpacingMarks(unittest.TestCase):
+    """Tests that spacing marks are indeed spacing."""
+
+    def setUp(self):
+        self.font_files, _ = load_fonts()
+        charset = coverage.character_set(self.font_files[0])
+        self.marks_to_test = [char for char in charset
+                              if unicode_data.category(char) in ['Lm', 'Sk']]
+        self.advance_cache = {}
+
+    def get_advances(self, text, font):
+        """Get a list of horizontal advances for text rendered in a font."""
+        try:
+            return self.advance_cache[(text, font)]
+        except KeyError:
+            hb_output = render.run_harfbuzz_on_text(text, font, '')
+            hb_output = json.loads(hb_output)
+            advances = [glyph['ax'] for glyph in hb_output]
+            self.advance_cache[(text, font)] = advances
+            return advances
+
+    def test_individual_spacing_marks(self):
+        """Tests that spacing marks are spacing by themselves."""
+        for font in self.font_files:
+            print 'Testing %s for stand-alone spacing marks...' % font
+            for mark in self.marks_to_test:
+                mark = unichr(mark)
+                advances = self.get_advances(mark, font)
+                assert len(advances) == 1
+                self.assertNotEqual(advances[0], 0)
+
+    def test_spacing_marks_in_combination(self):
+        """Tests that spacing marks do not combine with base letters."""
+        for font in self.font_files:
+            print 'Testing %s for spacing marks in combination...' % font
+            for base_letter in (u'A\u00C6BCDEFGHIJKLMNO\u00D8\u01A0PRST'
+                                u'U\u01AFVWXYZ'
+                                u'a\u00E6bcdefghi\u0131j\u0237klmn'
+                                u'o\u00F8\u01A1prs\u017Ftu\u01B0vwxyz'
+                                u'\u03D2'):
+                print 'Testing %s combinations' % base_letter
+                for mark in self.marks_to_test:
+                    mark = unichr(mark)
+                    advances = self.get_advances(base_letter + mark, font)
+                    self.assertEqual(len(advances), 2,
+                        'The sequence <%04X, %04X> combines, '
+                        'but it should not' % (ord(base_letter), ord(mark)))
+
 if __name__ == '__main__':
     unittest.main()
diff --git a/scripts/subset_for_web.py b/scripts/subset_for_web.py
index e100dfc..fa80590 100644
--- a/scripts/subset_for_web.py
+++ b/scripts/subset_for_web.py
@@ -1,4 +1,5 @@
 #!/usr/bin/python
+# coding=UTF-8
 #
 # Copyright 2014 Google Inc. All rights reserved.
 #
@@ -39,14 +40,58 @@ def read_charlist(filename):
             charlist.append(char)
         return charlist
 
+LATIN = (
+    range(0x0020, 0x007F) + range(0x00A0, 0x0100) +
+    [0x0131, 0x0152, 0x0153, 0x02C6, 0x02DA, 0x02DC, 0x2013, 0x2014, 0x2018,
+     0x2019, 0x201A, 0x201C, 0x201D, 0x201E, 0x2022, 0x2039, 0x203A, 0x2044,
+     0x2074, 0x20AC, 0x2212, 0x2215])
+
+CYRILLIC = range(0x0400, 0x0460) + [0x0490, 0x0491, 0x04B0, 0x04B1, 0x2116]
+
+SUBSETS = {
+    'cyrillic': LATIN + CYRILLIC,
+    'cyrillic-ext': (
+        LATIN + CYRILLIC + range(0x0460, 0x0530) + [0x20B4] +
+        range(0x2DE0, 0x2E00) + range(0xA640, 0xA6A0)),
+    'greek': LATIN + range(0x0384, 0x0400),
+    'greek-ext': LATIN + range(0x0384, 0x0400) + range(0x1F00, 0x2000),
+    'latin': LATIN,
+    'latin-ext': (
+        LATIN + range(0x0100, 0x0250) +
+        [0x02BC, 0x0300, 0x0301, 0x0303, 0x030F] +
+        range(0x1E00, 0x1F00) +
+        [0x2026, 0x2070, 0x2075, 0x2076, 0x2077, 0x2078, 0x2079, 0x207F] +
+        range(0x20A0, 0x20D0) + range(0x2C60, 0x2C80) +
+        range(0xA720, 0xA800)),
+    'menu': [ord(c) for c in u' ()DNQRabcfgortu΄ΕάαεηικλνКаилрцốữ'],
+}
+SUBSETS = {k: frozenset(v) for k, v in SUBSETS.iteritems()}
+
 
 def main(argv):
     """Subset the first argument to second, dropping unused parts of the font.
     """
-    charlist = read_charlist('res/web_charset.txt')
+    charlist = read_charlist('res/charsets/web.txt')
     # Add private use characters for legacy reasons
     charlist += [0xEE01, 0xEE02, 0xF6C3]
-    subset.subset_font(argv[1], argv[2], include=charlist)
+
+    features_to_keep = [
+        'c2sc', 'ccmp', 'cpsp', 'dlig', 'dnom', 'frac', 'kern', 'liga', 'lnum',
+        'locl', 'numr', 'onum', 'pnum', 'smcp', 'ss01', 'ss02', 'ss03', 'ss04',
+        'ss05', 'ss06', 'ss07', 'tnum']
+
+    source_filename = argv[1]
+    target_filename = argv[2]
+    subset.subset_font(
+        source_filename, target_filename,
+        include=charlist,
+        options={'layout_features': features_to_keep})
+
+    for suffix in SUBSETS.keys():
+        subset_filename = target_filename.replace('ttf', suffix)
+        subset.subset_font(
+            target_filename, subset_filename,
+            include=SUBSETS[suffix])
 
 
 if __name__ == '__main__':
diff --git a/scripts/touchup_for_web.py b/scripts/touchup_for_web.py
index 0f30728..b055d89 100755
--- a/scripts/touchup_for_web.py
+++ b/scripts/touchup_for_web.py
@@ -1,136 +1,17 @@
 #!/usr/bin/python
 """Post-build web fonts changes for Roboto."""
 
-import collections
 import os
 from os import path
 import sys
 
 from fontTools import ttLib
 from nototools import font_data
-from nototools import unicode_data
-
-
-def drop_lookup(table, lookup_number):
-    """Drop a lookup from an OpenType table by number.
-
-    Actually remove pointers from features to the lookup, which should be less
-    intrusive.
-    """
-    for feature in table.table.FeatureList.FeatureRecord:
-        if lookup_number in feature.Feature.LookupListIndex:
-            feature.Feature.LookupListIndex.remove(lookup_number)
-            feature.Feature.LookupCount -= 1
-
-
-def get_font_name(font):
-    """Gets the name of the font from the name table."""
-    return font_data.get_name_records(font)[4]
-
-
-DIGITS = ['zero', 'one', 'two', 'three', 'four',
-          'five', 'six', 'seven', 'eight', 'nine']
-
-def fix_digit_widths(font):
-    """Change all digit widths in the font to be the same."""
-    hmtx_table = font['hmtx']
-    widths = [hmtx_table[digit][0] for digit in DIGITS]
-    if len(set(widths)) > 1:
-        width_counter = collections.Counter(widths)
-        most_common_width = width_counter.most_common(1)[0][0]
-        print 'Digit widths were %s.' % repr(widths)
-        print 'Setting all glyph widths to %d.' % most_common_width
-        for digit in DIGITS:
-            assert abs(hmtx_table[digit][0] - most_common_width) <= 1
-            hmtx_table[digit][0] = most_common_width
-
-
-_MAP_SPACING_TO_COMBINING = {
-    'acute': 'acutecomb',
-    'breve': 'brevenosp',
-    'caron': 'uni030C',
-    'cedilla': 'cedillanosp',
-    'circumflex': 'circumflexnosp',
-    'dieresis': 'dieresisnosp',
-    'dotaccent': 'dotnosp',
-    'grave': 'gravecomb',
-    'hungarumlaut': 'acutedblnosp',
-    'macron': 'macroncomb',
-    'ogonek': 'ogoneknosp',
-    'tilde': 'tildecomb',
-    'ring': 'ringnosp',
-    'tonos': 'acutecomb',
-    'uni02F3': 'ringsubnosp',
-}
-
-def fix_ccmp_lookup(font):
-    """Fixes the broken ccmp lookup."""
-    cmap = font_data.get_cmap(font)
-    reverse_cmap = {name: code for (code, name) in cmap.items()}
-
-    # Where we know the bad 'ccmp' is
-    ccmp_lookup = font['GSUB'].table.LookupList.Lookup[2]
-    assert ccmp_lookup.LookupType == 4
-    assert ccmp_lookup.SubTableCount == 1
-    ligatures = ccmp_lookup.SubTable[0].ligatures
-    for first_char, ligtable in ligatures.iteritems():
-        ligatures_to_delete = []
-        for index, ligature in enumerate(ligtable):
-            assert len(ligature.Component) == 1
-            component = ligature.Component[0]
-            if (component.endswith('comb')
-                or component in ['commaaccent',
-                                 'commaaccentrotate',
-                                 'ringacute']):
-                continue
-            # https://code.google.com/a/google.com/p/roboto/issues/detail?id=54
-            if first_char == 'a' and component == 'uni02BE':
-                ligatures_to_delete.append(index)
-                continue
-            char = reverse_cmap[component]
-            general_category = unicode_data.category(char)
-            if general_category != 'Mn': # not a combining mark
-                ligature.Component[0] = _MAP_SPACING_TO_COMBINING[component]
-        ligatures[first_char] = [
-            ligature for (index, ligature) in enumerate(ligtable)
-            if index not in ligatures_to_delete]
 
 
 def apply_temporary_fixes(font):
     """Apply some temporary fixes.
     """
-    # Make sure macStyle is correct
-    # https://code.google.com/a/google.com/p/roboto/issues/detail?id=8
-    font_name = get_font_name(font)
-    bold = ('Bold' in font_name) or ('Black' in font_name)
-    italic = 'Italic' in font_name
-    font['head'].macStyle = (italic << 1) | bold
-
-    # Mark the font free for installation, embedding, etc.
-    # https://code.google.com/a/google.com/p/roboto/issues/detail?id=29
-    os2 = font['OS/2']
-    os2.fsType = 0
-
-    # Set the font vendor to Google
-    # https://code.google.com/a/google.com/p/roboto/issues/detail?id=46
-    os2.achVendID = 'GOOG'
-
-    # Drop the lookup forming the ff ligature
-    # https://code.google.com/a/google.com/p/roboto/issues/detail?id=47
-    drop_lookup(font['GSUB'], 5)
-
-    # Correct the ccmp lookup to use combining marks instead of spacing ones
-    # https://code.google.com/a/google.com/p/roboto/issues/detail?id=48
-    fix_ccmp_lookup(font)
-
-    # Fix the digit widths
-    # https://code.google.com/a/google.com/p/roboto/issues/detail?id=49
-    fix_digit_widths(font)
-
-    # Add cmap for U+2117 SOUND RECORDING COPYRIGHT
-    # https://code.google.com/a/google.com/p/roboto/issues/detail?id=44
-    font_data.add_to_cmap(font, {0x2117: 'published'})
-
     # Fix version number from buildnumber.txt
     # https://code.google.com/a/google.com/p/roboto/issues/detail?id=50
     from datetime import date
@@ -139,17 +20,8 @@ def apply_temporary_fixes(font):
         path.dirname(__file__), os.pardir, 'res', 'buildnumber.txt')
     build_number = open(build_number_txt).read().strip()
 
-    version_record = 'Version 2.0%s; %d' % (build_number, date.today().year)
-
-    for record in font['name'].names:
-        if record.nameID == 5:
-            if record.platformID == 1 and record.platEncID == 0:  # MacRoman
-                record.string = version_record
-            elif record.platformID == 3 and record.platEncID == 1:
-                # Windows UCS-2
-                record.string = version_record.encode('UTF-16BE')
-            else:
-                assert False
+    version_record = 'Version 2.%s; %d' % (build_number, date.today().year)
+    font_data.set_name_record(font, 5, version_record)
 
 
 def apply_web_specific_fixes(font):
@@ -167,11 +39,40 @@ def apply_web_specific_fixes(font):
     os2.usWinAscent = 1946
     os2.usWinDescent = 512
 
+    family_name = 'RobotoDraft'
+    subfamily_name = font_data.get_name_records(font)[2].encode('ASCII')
+    assert(subfamily_name in
+        ['Thin', 'Thin Italic',
+         'Light', 'Light Italic',
+         'Regular', 'Italic',
+         'Medium', 'Medium Italic',
+         'Bold', 'Bold Italic',
+         'Black', 'Black Italic'])
+    full_name = family_name + ' ' + subfamily_name
+    year = '2014'
+
+    # Copyright message
+    font_data.set_name_record(
+        font, 0, 'Copyright %s Google Inc. All Rights Reserved.' % year)
+
+    # Family name
+    font_data.set_name_record(font, 1, family_name)
+    font_data.set_name_record(font, 16, family_name)
+
+    # Unique identifier and full name
+    font_data.set_name_record(font, 3, full_name)
+    font_data.set_name_record(font, 4, full_name)
+    font_data.set_name_record(font, 18, None)
+
+    # PostScript name
+    font_data.set_name_record(
+        font, 6, family_name+'-'+subfamily_name.replace(' ', ''))
+
 
 def correct_font(source_font_name, target_font_name):
     """Corrects metrics and other meta information."""
     font = ttLib.TTFont(source_font_name)
-#    apply_temporary_fixes(font)
+    apply_temporary_fixes(font)
     apply_web_specific_fixes(font)
     font.save(target_font_name)