third_party/harfbuzz/contrib/tables/grapheme-break-parse.py

   1 import sys
   2 from unicode_parse_common import *
   3
   4 # http://www.unicode.org/Public/UNIDATA/auxiliary/GraphemeBreakProperty.txt
   5
   6 property_to_harfbuzz = {
   7   'CR': 'HB_Grapheme_CR',
   8   'LF': 'HB_Grapheme_LF',
   9   'Control': 'HB_Grapheme_Control',
  10   'Extend': 'HB_Grapheme_Extend',
  11   'Prepend': 'HB_Grapheme_Other',
  12   'SpacingMark': 'HB_Grapheme_Other',
  13   'L': 'HB_Grapheme_L',
  14   'V': 'HB_Grapheme_V',
  15   'T': 'HB_Grapheme_T',
  16   'LV': 'HB_Grapheme_LV',
  17   'LVT': 'HB_Grapheme_LVT',
  18 }
  19
  20 def main(infile, outfile):
  21   ranges = unicode_file_parse(infile, property_to_harfbuzz)
  22   ranges.sort()
  23
  24   print >>outfile, '// Generated from Unicode Grapheme break tables\n'
  25   print >>outfile, '#ifndef GRAPHEME_BREAK_PROPERTY_H_'
  26   print >>outfile, '#define GRAPHEME_BREAK_PROPERTY_H_\n'
  27   print >>outfile, '#include <stdint.h>'
  28   print >>outfile, '#include "harfbuzz-external.h"\n'
  29   print >>outfile, 'struct grapheme_break_property {'
  30   print >>outfile, '  uint32_t range_start;'
  31   print >>outfile, '  uint32_t range_end;'
  32   print >>outfile, '  HB_GraphemeClass klass;'
  33   print >>outfile, '};\n'
  34   print >>outfile, 'static const struct grapheme_break_property grapheme_break_properties[] = {'
  35   for (start, end, value) in ranges:
  36     print >>outfile, '  {0x%x, 0x%x, %s},' % (start, end, value)
  37   print >>outfile, '};\n'
  38   print >>outfile, 'static const unsigned grapheme_break_properties_count = %d;\n' % len(ranges)
  39   print >>outfile, '#endif  // GRAPHEME_BREAK_PROPERTY_H_'
  40
  41 if __name__ == '__main__':
  42   if len(sys.argv) != 3:
  43     print 'Usage: %s <input .txt> <output .h>' % sys.argv[0]
  44   else:
  45     main(file(sys.argv[1], 'r'), file(sys.argv[2], 'w+'))