diff options
author | Stephen Dolan <mu@netsoc.tcd.ie> | 2013-09-14 19:30:39 +0100 |
---|---|---|
committer | Stephen Dolan <mu@netsoc.tcd.ie> | 2013-09-14 19:30:39 +0100 |
commit | dae2422fd18487c89dd79dc54eb4345861ab7bdc (patch) | |
tree | 14f046128de9c35a57e3574a3a4365095b680d4f /scripts | |
parent | f98e3e6c26b3b53701e1e6c2ff98ab6227cde3b8 (diff) | |
parent | 37cfc912c1f384d177162f8aa706452754d2c6ab (diff) |
Merge branch 'libjq'
Conflicts:
Makefile.am
configure.ac
Diffstat (limited to 'scripts')
-rw-r--r-- | scripts/gen_utf8_tables.py | 32 |
1 files changed, 32 insertions, 0 deletions
diff --git a/scripts/gen_utf8_tables.py b/scripts/gen_utf8_tables.py new file mode 100644 index 00000000..65d03a2b --- /dev/null +++ b/scripts/gen_utf8_tables.py @@ -0,0 +1,32 @@ +#!/usr/bin/python + +# This program was used to generate jv_utf8_tables.gen.h + +mask = lambda n: (1 << n) - 1 + + +def print_table(type, name, t): + print("static const %s %s[] =" % (type, name)) + first = True + for i in range(0,len(t),16): + print ((" {" if i == 0 else " ") + + ", ".join("0x%02x"%n for n in t[i:i+16]) + + ("," if i + 16 < len(t) else "};")) + + +def utf8info(c): + if c < 0x80: return 1, mask(7) + if 0x80 <= c <= 0xBF: return 255, mask(6) + if 0xC0 <= c <= 0xC1: return 0, 0 + if 0xC2 <= c <= 0xDF: return 2, mask(5) + if 0xE0 <= c <= 0xEF: return 3, mask(4) + if 0xF0 <= c <= 0xF4: return 4, mask(3) + if 0xF4 <= c <= 0xFF: return 0, 0 + +table = lambda i: [utf8info(c)[i] for c in range(256)] + +print("#define UTF8_CONTINUATION_BYTE ((unsigned char)255)") + +print_table("unsigned char", "utf8_coding_length", table(0)) +print_table("unsigned char", "utf8_coding_bits", table(1)) +print_table("int", "utf8_first_codepoint", [0, 0x0, 0x80, 0x800, 0x10000]) |