source: proto/charsetcompiler/inputs/UTF8_XML @ 3903

Last change on this file since 3903 was 1201, checked in by cameron, 8 years ago

Xerces prototype files

File size: 665 bytes
Line 
1#  Definitions for UTF-8 plus XML 0xFFFE/0xFFFF identification.
2#
3u8_unibyte = [\x00-\x7F]
4u8_prefix = [\xC0-\xFF]
5u8_suffix = [\x80-\xBF]
6# 2-byte sequences
7u8_prefix2 = [\xC0-\xDF]
8# For 2 byte-sequence validation
9u8_badprefix2 = [\xC0-\xC1]
10# 3-byte sequences
11u8_prefix3or4 = [\xE0-\xFF]
12u8_prefix3 = [\xE0-\xEF]
13# For 3 byte-sequence validation
14xE0 = [\xE0]
15xED = [\xED]
16xA0_xBF = [\xA0-\xBF]
17x80_x9F = [\x80-\x9F]
18# For FFFE/FFFF recognition in XML
19xEF = [\xEF]
20xBF = [\xBF]
21xBE = [\xBE]
22# 4-byte sequences
23u8_prefix4 = [\xF0-\xFF]
24# 4 byte sequence validation
25u8_badprefix4 = [\xF5-\xFF]
26xF0 = [\xF0]
27xF4 = [\xF4]
28x90_xBF = [\x90-\xBF]
29x80_x8F = [\x80-\x8F]
30
Note: See TracBrowser for help on using the repository browser.