diff options
author | Roozbeh Pournader <roozbeh@google.com> | 2015-01-09 20:05:20 -0800 |
---|---|---|
committer | James Godfrey-Kittle <jamesgk@google.com> | 2015-04-16 12:16:25 -0700 |
commit | 7b0eff6b32c6aec1c2e59ffae8bf02dc47bcd942 (patch) | |
tree | aa5fa35eb6bdf6baa0e547fb94148b908cfcfdf5 /scripts | |
parent | 8c98e08586f15efbabca2a9fc8e501677a4d8e89 (diff) |
Update subsets for web.
Diffstat (limited to 'scripts')
-rwxr-xr-x | scripts/subset_for_web.py | 27 |
1 files changed, 18 insertions, 9 deletions
diff --git a/scripts/subset_for_web.py b/scripts/subset_for_web.py index 57dc8ce..574df58 100755 --- a/scripts/subset_for_web.py +++ b/scripts/subset_for_web.py @@ -46,25 +46,34 @@ LATIN = ( 0x2019, 0x201A, 0x201C, 0x201D, 0x201E, 0x2022, 0x2039, 0x203A, 0x2044, 0x2074, 0x20AC, 0x2212, 0x2215]) -CYRILLIC = range(0x0400, 0x0460) + [0x0490, 0x0491, 0x04B0, 0x04B1, 0x2116] +CYRILLIC = ( + range(0x0400, 0x0460) + + [0x0490, 0x0491, 0x04B0, 0x04B1, 0x20BD, 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), + 'greek': LATIN + range(0x0370, 0x0400), + 'greek-ext': LATIN + range(0x0384, 0x0400) + range(0x1F00, 0x2000), 'latin': LATIN, 'latin-ext': ( - LATIN + range(0x0100, 0x0250) + + LATIN + range(0x0100, 0x0370) + [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'], + range(0x1D00, 0x1F00) + + [0x2026] + + range(0x2070, 0x20D0) + + range(0x2C60, 0x2C80) + + range(0xA700, 0xA800)), + 'menu': [ord(c) for c in u' ()DEKNQRabcfgoprtuvĸ΄ΕάαεηικλνКаилрцốữ'], + 'vietnamese': ( + LATIN + + [0x0102, 0x0103, 0x0110, 0x0111, 0x0128, 0x0129, 0x0168, 0x0169, + 0x01A0, 0x01A1, 0x01AF, 0x01B0, 0x02D8, 0x0309, 0x0323] + + range(0x1EA0, 0x1EFA) + [0x20AB]), } + SUBSETS = {k: frozenset(v) for k, v in SUBSETS.iteritems()} |