| 4 | import sys |
| 5 | |
| 6 | def main(input, output): |
| 7 | code_point_re = re.compile(r'^U\+([0-9a-fA-F]+)\s*;\s*ExtendedPictographic.*$') |
| 8 | code_point_range_re = re.compile(r'^U\+([0-9a-fA-F]+)\.\.U\+([0-9a-fA-F]+)\s*;\s*ExtendedPictographic.*$') |
| 9 | |
| 10 | for line in input: |
| 11 | m = code_point_re.match(line) |
| 12 | if m: |
| 13 | print('set.add(0x' + m.group(1) + ');', file=output) |
| 14 | else: |
| 15 | m = code_point_range_re.match(line) |
| 16 | if m: |
| 17 | print('set.add(0x' + m.group(1) + ', 0x' + m.group(2) + ');', file=output) |
| 18 | |
| 19 | if __name__ == '__main__': |
| 20 | with codecs.open(sys.argv[1], 'r', 'utf-8') as f: |