third_party/harfbuzz/contrib/tables/category-parse.py

   1 import sys
   2 from unicode_parse_common import *
   3
   4 # http://www.unicode.org/Public/5.1.0/ucd/extracted/DerivedGeneralCategory.txt
   5
   6 category_to_harfbuzz = {
   7   'Mn': 'HB_Mark_NonSpacing',
   8   'Mc': 'HB_Mark_SpacingCombining',
   9   'Me': 'HB_Mark_Enclosing',
  10
  11   'Nd': 'HB_Number_DecimalDigit',
  12   'Nl': 'HB_Number_Letter',
  13   'No': 'HB_Number_Other',
  14
  15   'Zs': 'HB_Separator_Space',
  16   'Zl': 'HB_Separator_Line',
  17   'Zp': 'HB_Separator_Paragraph',
  18
  19   'Cc': 'HB_Other_Control',
  20   'Cf': 'HB_Other_Format',
  21   'Cs': 'HB_Other_Surrogate',
  22   'Co': 'HB_Other_PrivateUse',
  23   'Cn': 'HB_Other_NotAssigned',
  24
  25   'Lu': 'HB_Letter_Uppercase',
  26   'Ll': 'HB_Letter_Lowercase',
  27   'Lt': 'HB_Letter_Titlecase',
  28   'Lm': 'HB_Letter_Modifier',
  29   'Lo': 'HB_Letter_Other',
  30
  31   'Pc': 'HB_Punctuation_Connector',
  32   'Pd': 'HB_Punctuation_Dash',
  33   'Ps': 'HB_Punctuation_Open',
  34   'Pe': 'HB_Punctuation_Close',
  35   'Pi': 'HB_Punctuation_InitialQuote',
  36   'Pf': 'HB_Punctuation_FinalQuote',
  37   'Po': 'HB_Punctuation_Other',
  38
  39   'Sm': 'HB_Symbol_Math',
  40   'Sc': 'HB_Symbol_Currency',
  41   'Sk': 'HB_Symbol_Modifier',
  42   'So': 'HB_Symbol_Other',
  43 }
  44
  45 def main(infile, outfile):
  46   ranges = unicode_file_parse(infile, category_to_harfbuzz)
  47   ranges = sort_and_merge(ranges)
  48
  49   print >>outfile, '// Generated from Unicode script tables\n'
  50   print >>outfile, '#ifndef CATEGORY_PROPERTIES_H_'
  51   print >>outfile, '#define CATEGORY_PROPERTIES_H_\n'
  52   print >>outfile, '#include <stdint.h>'
  53   print >>outfile, '#include "harfbuzz-external.h"\n'
  54   print >>outfile, 'struct category_property {'
  55   print >>outfile, '  uint32_t range_start;'
  56   print >>outfile, '  uint32_t range_end;'
  57   print >>outfile, '  HB_CharCategory category;'
  58   print >>outfile, '};\n'
  59   print >>outfile, 'static const struct category_property category_properties[] = {'
  60   for (start, end, value) in ranges:
  61     print >>outfile, '  {0x%x, 0x%x, %s},' % (start, end, value)
  62   print >>outfile, '};\n'
  63   print >>outfile, 'static const unsigned category_properties_count = %d;\n' % len(ranges)
  64   print >>outfile, '#endif  // CATEGORY_PROPERTIES_H_'
  65
  66 if __name__ == '__main__':
  67   if len(sys.argv) != 3:
  68     print 'Usage: %s <input .txt> <output .h>' % sys.argv[0]
  69   else:
  70     main(file(sys.argv[1], 'r'), file(sys.argv[2], 'w+'))