Files
hawk/tools/gen-jis0208-tabs.hawk
hyung-hwan a758747661
All checks were successful
continuous-integration/drone/push Build is passing
added the jis0208 cmgr
2026-04-11 00:38:11 +09:00

237 lines
5.0 KiB
Plaintext

# Generate the dense JIS X 0208 lookup arrays used by lib/jis0208.c
#
# Usage:
# hawk -f tools/gen-jis0208-tabs.hawk -- tools/JIS0208.TXT
# hawk -f tools/gen-jis0208-tabs.hawk -- --forward tools/JIS0208.TXT
# hawk -f tools/gen-jis0208-tabs.hawk -- --reverse tools/JIS0208.TXT
#
BEGIN {
MODE = "all";
bad = 0;
init_forward_pages();
init_reverse_pages();
parse_args();
}
/^[ \t]*0x[0-9A-Fa-f]+[ \t]+0x[0-9A-Fa-f]+[ \t]+0x[0-9A-Fa-f]+/ {
@local jis, uc;
jis = parse_hex($2);
uc = parse_hex($3);
if (jis < 0 || uc < 0) next;
add_forward_mapping(jis, uc);
add_reverse_mapping(jis, uc);
}
END {
@local i, first;
if (bad) exit 1;
printf("/* generated by tools/gen-jis0208-tabs.hawk from %s */\n\n", input_name());
first = 1;
if (MODE == "all" || MODE == "forward")
{
for (i = 1; i <= fwd_page_count; i++)
{
if (!first) printf("\n");
dump_forward_page(i);
first = 0;
}
}
if (MODE == "all" || MODE == "reverse")
{
for (i = 1; i <= rev_page_count; i++)
{
if (!first) printf("\n");
dump_reverse_page(i);
first = 0;
}
}
}
function parse_args()
{
@local i, a;
for (i = 1; i < ARGC; i++)
{
a = ARGV[i];
if (a == "" || substr(a, 1, 2) != "--") continue;
if (a == "--all")
MODE = "all";
else if (a == "--forward")
MODE = "forward";
else if (a == "--reverse")
MODE = "reverse";
else
{
printf("unknown option %s\n", a) > "/dev/stderr";
bad = 1;
}
ARGV[i] = "";
}
}
function input_name()
{
@local i, sep, out;
sep = "";
out = "";
for (i = 1; i < ARGC; i++)
{
if (ARGV[i] == "") continue;
out = out sep ARGV[i];
sep = ", ";
}
return (out != "")? out: "<stdin>";
}
function hex_digit(ch)
{
if (ch >= "0" && ch <= "9") return ch + 0;
if (ch >= "A" && ch <= "F") return 10 + index("ABCDEF", ch) - 1;
if (ch >= "a" && ch <= "f") return 10 + index("abcdef", ch) - 1;
return -1;
}
function parse_hex(s)
{
@local i, n, d;
n = 0;
if (substr(s, 1, 2) == "0x" || substr(s, 1, 2) == "0X")
s = substr(s, 3);
for (i = 1; i <= length(s); i++)
{
d = hex_digit(substr(s, i, 1));
if (d < 0) return -1;
n = n * 16 + d;
}
return n;
}
function init_forward_pages()
{
fwd_page_count = 1;
fwd_page_no[1] = 0;
fwd_name[1] = "jis0208_to_uc_0";
fwd_from[1] = 0x2121;
fwd_to[1] = 0x7426;
fwd_len[1] = fwd_to[1] - fwd_from[1] + 1;
}
function init_reverse_pages()
{
rev_page_count = 10;
rev_page_no[1] = 0; rev_name[1] = "uc_to_jis0208_0"; rev_from[1] = 0x00A1; rev_to[1] = 0x00FF;
rev_page_no[2] = 1; rev_name[2] = "uc_to_jis0208_1"; rev_from[2] = 0x0391; rev_to[2] = 0x0451;
rev_page_no[3] = 2; rev_name[3] = "uc_to_jis0208_2"; rev_from[3] = 0x2010; rev_to[3] = 0x2FFF;
rev_page_no[4] = 3; rev_name[4] = "uc_to_jis0208_3"; rev_from[4] = 0x3000; rev_to[4] = 0x3FFF;
rev_page_no[5] = 4; rev_name[5] = "uc_to_jis0208_4"; rev_from[5] = 0x4E00; rev_to[5] = 0x5FFF;
rev_page_no[6] = 5; rev_name[6] = "uc_to_jis0208_5"; rev_from[6] = 0x6000; rev_to[6] = 0x6FFF;
rev_page_no[7] = 6; rev_name[7] = "uc_to_jis0208_6"; rev_from[7] = 0x7000; rev_to[7] = 0x7FFF;
rev_page_no[8] = 7; rev_name[8] = "uc_to_jis0208_7"; rev_from[8] = 0x8000; rev_to[8] = 0x8FFF;
rev_page_no[9] = 8; rev_name[9] = "uc_to_jis0208_8"; rev_from[9] = 0x9000; rev_to[9] = 0x9FA0;
rev_page_no[10] = 9; rev_name[10] = "uc_to_jis0208_9"; rev_from[10] = 0xFF01; rev_to[10] = 0xFFE5;
for (i = 1; i <= rev_page_count; i++)
rev_len[i] = rev_to[i] - rev_from[i] + 1;
}
function add_forward_mapping(jis, uc)
{
@local i, idx, old;
for (i = 1; i <= fwd_page_count; i++)
{
if (jis < fwd_from[i] || jis > fwd_to[i]) continue;
idx = jis - fwd_from[i];
old = fwd_data[i][idx] + 0;
if (old != 0 && old != uc)
{
printf("duplicate forward mapping for 0x%04x: 0x%04x vs 0x%04x\n", jis, old, uc) > "/dev/stderr";
bad = 1;
return;
}
fwd_data[i][idx] = uc;
return;
}
}
function add_reverse_mapping(jis, uc)
{
@local i, idx, old;
for (i = 1; i <= rev_page_count; i++)
{
if (uc < rev_from[i] || uc > rev_to[i]) continue;
idx = uc - rev_from[i];
old = rev_data[i][idx] + 0;
if (old != 0 && old != jis)
{
printf("duplicate reverse mapping for U+%04x: 0x%04x vs 0x%04x\n", uc, old, jis) > "/dev/stderr";
bad = 1;
return;
}
rev_data[i][idx] = jis;
return;
}
}
function dump_dense_array(kind, page, total)
{
@local i, v;
for (i = 0; i < total; i++)
{
if ((i % 9) == 0) printf("\t");
if (kind == "forward")
v = fwd_data[page][i] + 0;
else
v = rev_data[page][i] + 0;
printf("0x%04x", v);
if (i + 1 < total) printf(", ");
if ((i % 9) == 8 || i + 1 >= total)
printf("\n");
}
}
function dump_forward_page(i)
{
printf("/* page %d 0x%04x-0x%04x */\n", fwd_page_no[i], fwd_from[i], fwd_to[i]);
printf("static const hawk_uint16_t %s[] = {\n", fwd_name[i]);
dump_dense_array("forward", i, fwd_len[i]);
printf("};\n");
}
function dump_reverse_page(i)
{
printf("/* page %d 0x%04x-0x%04x */\n", rev_page_no[i], rev_from[i], rev_to[i]);
printf("static const hawk_uint16_t %s[] = {\n", rev_name[i]);
dump_dense_array("reverse", i, rev_len[i]);
printf("};\n");
}