1 import sys 2 from unicode_parse_common import * 3 4 # http://www.unicode.org/Public/5.1.0/ucd/Scripts.txt 5 6 script_to_harfbuzz = { 7 # This is the list of HB_Script_* at the time of writing 8 'Common': 'HB_Script_Common', 9 'Greek': 'HB_Script_Greek', 10 'Cyrillic': 'HB_Script_Cyrillic', 11 'Armenian': 'HB_Script_Armenian', 12 'Hebrew': 'HB_Script_Hebrew', 13 'Arabic': 'HB_Script_Arabic', 14 'Syriac': 'HB_Script_Syriac', 15 'Thaana': 'HB_Script_Thaana', 16 'Devanagari': 'HB_Script_Devanagari', 17 'Bengali': 'HB_Script_Bengali', 18 'Gurmukhi': 'HB_Script_Gurmukhi', 19 'Gujarati': 'HB_Script_Gujarati', 20 'Oriya': 'HB_Script_Oriya', 21 'Tamil': 'HB_Script_Tamil', 22 'Telugu': 'HB_Script_Telugu', 23 'Kannada': 'HB_Script_Kannada', 24 'Malayalam': 'HB_Script_Malayalam', 25 'Sinhala': 'HB_Script_Sinhala', 26 'Thai': 'HB_Script_Thai', 27 'Lao': 'HB_Script_Lao', 28 'Tibetan': 'HB_Script_Tibetan', 29 'Myanmar': 'HB_Script_Myanmar', 30 'Georgian': 'HB_Script_Georgian', 31 'Hangul': 'HB_Script_Hangul', 32 'Ogham': 'HB_Script_Ogham', 33 'Runic': 'HB_Script_Runic', 34 'Khmer': 'HB_Script_Khmer', 35 'Inherited': 'HB_Script_Inherited', 36 } 37 38 class ScriptDict(object): 39 def __init__(self, base): 40 self.base = base 41 42 def __getitem__(self, key): 43 r = self.base.get(key, None) 44 if r is None: 45 return 'HB_Script_Common' 46 return r 47 48 def main(infile, outfile): 49 ranges = unicode_file_parse(infile, 50 ScriptDict(script_to_harfbuzz), 51 'HB_Script_Common') 52 ranges = sort_and_merge(ranges) 53 54 print >>outfile, '// Generated from Unicode script tables\n' 55 print >>outfile, '#ifndef SCRIPT_PROPERTIES_H_' 56 print >>outfile, '#define SCRIPT_PROPERTIES_H_\n' 57 print >>outfile, '#include <stdint.h>' 58 print >>outfile, '#include "harfbuzz-shaper.h"\n' 59 print >>outfile, 'struct script_property {' 60 print >>outfile, ' uint32_t range_start;' 61 print >>outfile, ' uint32_t range_end;' 62 print >>outfile, ' HB_Script script;' 63 print >>outfile, '};\n' 64 print >>outfile, 'static const struct script_property script_properties[] = {' 65 for (start, end, value) in ranges: 66 print >>outfile, ' {0x%x, 0x%x, %s},' % (start, end, value) 67 print >>outfile, '};\n' 68 print >>outfile, 'static const unsigned script_properties_count = %d;\n' % len(ranges) 69 print >>outfile, '#endif // SCRIPT_PROPERTIES_H_' 70 71 if __name__ == '__main__': 72 if len(sys.argv) != 3: 73 print 'Usage: %s <input .txt> <output .h>' % sys.argv[0] 74 else: 75 main(file(sys.argv[1], 'r'), file(sys.argv[2], 'w+')) 76
